This PR adds official support for Haitian Creole (ht) to spaCy's spacy/lang module.
It includes:
Added all core language data files for spacy/lang/ht:
tokenizer_exceptions.py
punctuation.py
lex_attrs.py
syntax_iterators.py
lemmatizer.py
stop_words.py
tag_map.py
Unit tests for tokenizer and noun chunking (test_tokenizer.py, test_noun_chunking.py, etc.). Passed all 58 pytest spacy/tests/lang/ht tests that I've created.
Basic tokenizer rules adapted for Haitian Creole orthography and informal contractions.
Custom like_num atrribute supporting Haitian number formats (e.g., "3yèm").
Support for common informal apostrophe usage (e.g., "m'ap", "n'ap", "di'm").
Ensured no breakages in other language modules.
Followed spaCy coding style (PEP8, Black).
This provides a foundation for Haitian Creole NLP development using spaCy.
* Correct code example for Span.lemma_ in API Docs (#13405)
* Correct documented return type of Vocab.to_bytes in API docs
* Correct wording for Vectors.__init__ in API docs
This PR removes the dependency on langcodes introduced in #9342.
While the introduction of langcodes allows a significantly wider range of language codes, there are some unexpected side effects:
zh-Hant (Traditional Chinese) should be mapped to zh intead of None, as spaCy's Chinese model is based on pkuseg which supports tokenization of both Simplified and Traditional Chinese.
Since it is possible that spaCy may have a model for Norwegian Nynorsk in the future, mapping no (macrolanguage Norwegian) to nb (Norwegian Bokmål) might be misleading. In that case, the user should be asked to specify nb or nn (Norwegian Nynorsk) specifically or consult the doc.
Same as above for regional variants of languages such as en_gb and en_us.
Overall, IMHO, introducing an extra dependency just for the conversion of language codes is an overkill. It is possible that most user just need the conversion between 2/3-letter ISO codes and a simple dictionary lookup should suffice.
With this PR, ISO 639-1 and ISO 639-3 codes are supported. ISO 639-2/B (bibliographic codes which are not favored and used in ISO 639-3) and deprecated ISO 639-1/2 codes are also supported to maximize backward compatibility.
After the Cython 3 change, the types of enum members such as
spacy.parts_of_speech.NOUN became 'flag', rather than simple 'int'.
This change mostly doesn't matter because the flag type does duck-type
like an int -- it compares, additions, prints etc the same. However,
it doesn't repr the same and if you do an isinstance check it will fail.
It's therefore better to just make them ints like they were before.
In order to support Python 3.13, we had to migrate to Cython 3.0. This caused some tricky interaction with our Pydantic usage, because Cython 3 uses the from __future__ import annotations semantics, which causes type annotations to be saved as strings.
The end result is that we can't have Language.factory decorated functions in Cython modules anymore, as the Language.factory decorator expects to inspect the signature of the functions and build a Pydantic model. If the function is implemented in Cython, an error is raised because the type is not resolved.
To address this I've moved the factory functions into a new module, spacy.pipeline.factories. I've added __getattr__ importlib hooks to the previous locations, in case anyone was importing these functions directly. The change should have no backwards compatibility implications.
Along the way I've also refactored the registration of functions for the config. Previously these ran as import-time side-effects, using the registry decorator. I've created instead a new module spacy.registrations. When the registry is accessed it calls a function ensure_populated(), which cases the registrations to occur.
I've made a similar change to the Language.factory registrations in the new spacy.pipeline.factories module.
I want to remove these import-time side-effects so that we can speed up the loading time of the library, which can be especially painful on the CLI. I also find that I'm often working to track down the implementations of functions referenced by strings in the config. Having the registrations all happen in one place will make this easier.
With these changes I've fortunately avoided the need to migrate to Pydantic v2 properly --- we're still using the v1 compatibility shim. We might not be able to hold out forever though: Pydantic (reasonably) aren't actively supporting the v1 shims. I put a lot of work into v2 migration when investigating the 3.13 support, and it's definitely challenging. In any case, it's a relief that we don't have to do the v2 migration at the same time as the Cython 3.0/Python 3.13 support.
* Fix bug in memory-zone code when adding non-transient strings. The error could result in segmentation faults or other memory errors during memory zones if new labels were added to the model.
* Fix handling of new morphological labels within memory zones. Addresses second issue reported in Memory leak of MorphAnalysis object. #13684
Add a context manage nlp.memory_zone(), which will begin
memory_zone() blocks on the vocab, string store, and potentially
other components.
Example usage:
```
with nlp.memory_zone():
for text in nlp.pipe(texts):
do_something(doc)
# do_something(doc) <-- Invalid
```
Once the memory_zone() block expires, spaCy will free any shared
resources that were allocated for the text-processing that occurred
within the memory_zone. If you create Doc objects within a memory
zone, it's invalid to access them once the memory zone is expired.
The purpose of this is that spaCy creates and stores Lexeme objects
in the Vocab that can be shared between multiple Doc objects. It also
interns strings. Normally, spaCy can't know when all Doc objects using
a Lexeme are out-of-scope, so new Lexemes accumulate in the vocab,
causing memory pressure.
Memory zones solve this problem by telling spaCy "okay none of the
documents allocated within this block will be accessed again". This
lets spaCy free all new Lexeme objects and other data that were
created during the block.
The mechanism is general, so memory_zone() context managers can be
added to other components that could benefit from them, e.g. pipeline
components.
I experimented with adding memory zone support to the tokenizer as well,
for its cache. However, this seems unnecessarily complicated. It makes
more sense to just stick a limit on the cache size. This lets spaCy
benefit from the efficiency advantage of the cache better, because
we can maintain a (bounded) cache even if only small batches of
documents are being processed.
* Add workflow files for cibuildwheel
* Add config for cibuildwheel
* Set version for experimental prerelease
* Try updating cython
* Skip 32-bit windows builds
* Revert "Try updating cython"
This reverts commit c1b794ab5c.
* Try to import cibuildwheel settings from previous setup
* fix type annotation in docs
* only restore entities after loss calculation
* restore entities of sample in initialization
* rename overfitting function
* fix EL scorer
* Relax test
* fix formatting
* Update spacy/pipeline/entity_linker.py
Co-authored-by: Raphael Mitsch <r.mitsch@outlook.com>
* rename to _ensure_ents
* further rename
* allow for scorer to be None
---------
Co-authored-by: Raphael Mitsch <r.mitsch@outlook.com>
The 'direct' option in 'spacy download' is supposed to only download from our model releases repository. However, users were able to pass in a relative path, allowing download from arbitrary repositories. This meant that a service that sourced strings from user input and which used the direct option would allow users to install arbitrary packages.
* TextCatParametricAttention.v1: set key transform dimensions
This is necessary for tok2vec implementations that initialize
lazily (e.g. curated transformers).
* Add lazily-initialized tok2vec to simulate transformers
Add a lazily-initialized tok2vec to the tests and test the current
textcat models with it.
Fix some additional issues found using this test.
* isort
* Add `test.` prefix to `LazyInitTok2Vec.v1`
The doc/token extension serialization tests add extensions that are not
serializable with pickle. This didn't cause issues before due to the
implicit run order of tests. However, test ordering has changed with
pytest 8.0.0, leading to failed tests in test_language.
Update the fixtures in the extension serialization tests to do proper
teardown and remove the extensions.
macOS now uses port 5000 for the AirPlay receiver functionality, so this
test will always fail on a macOS desktop (unless AirPlay receiver
functionality is disabled like in CI).
Before this change, the workers of pipe call with n_process != 1 were
stopped by calling `terminate` on the processes. However, terminating a
process can leave queues, pipes, and other concurrent data structures in
an invalid state.
With this change, we stop using terminate and take the following approach
instead:
* When the all documents are processed, the parent process puts a
sentinel in the queue of each worker.
* The parent process then calls `join` on each worker process to
let them finish up gracefully.
* Worker processes break from the queue processing loop when the
sentinel is encountered, so that they exit.
We need special handling when one of the workers encounters an error and
the error handler is set to raise an exception. In this case, we cannot
rely on the sentinel to finish all workers -- the queue is a FIFO queue
and there may be other work queued up before the sentinel. We use the
following approach to handle error scenarios:
* The parent puts the end-of-work sentinel in the queue of each worker.
* The parent closes the reading-end of the channel of each worker.
* Then:
- If the worker was waiting for work, it will encounter the sentinel
and break from the processing loop.
- If the worker was processing a batch, it will attempt to write
results to the channel. This will fail because the channel was
closed by the parent and the worker will break from the processing
loop.
* Add spacy.TextCatParametricAttention.v1
This layer provides is a simplification of the ensemble classifier that
only uses paramteric attention. We have found empirically that with a
sufficient amount of training data, using the ensemble classifier with
BoW does not provide significant improvement in classifier accuracy.
However, plugging in a BoW classifier does reduce GPU training and
inference performance substantially, since it uses a GPU-only kernel.
* Fix merge fallout
* Add TextCatReduce.v1
This is a textcat classifier that pools the vectors generated by a
tok2vec implementation and then applies a classifier to the pooled
representation. Three reductions are supported for pooling: first, max,
and mean. When multiple reductions are enabled, the reductions are
concatenated before providing them to the classification layer.
This model is a generalization of the TextCatCNN model, which only
supports mean reductions and is a bit of a misnomer, because it can also
be used with transformers. This change also reimplements TextCatCNN.v2
using the new TextCatReduce.v1 layer.
* Doc fixes
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Fully specify `TextCatCNN` <-> `TextCatReduce` equivalence
* Move TextCatCNN docs to legacy, in prep for moving to spacy-legacy
* Add back a test for TextCatCNN.v2
* Replace TextCatCNN in pipe configurations and templates
* Add an infobox to the `TextCatReduce` section with an `TextCatCNN` anchor
* Add last reduction (`use_reduce_last`)
* Remove non-working TextCatCNN Netlify redirect
* Revert layer changes for the quickstart
* Revert one more quickstart change
* Remove unused import
* Fix docstring
* Fix setting name in error message
---------
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update README.md to include links for GPU processing, LLM, and spaCy's blog.
* Create ojo4f3.md
* corrected README to most current version with links to GPU processing, LLM's, and the spaCy blog.
* Delete .github/contributors/ojo4f3.md
* changed LLM icon
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Apply suggestions from code review
---------
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update `TextCatBOW` to use the fixed `SparseLinear` layer
A while ago, we fixed the `SparseLinear` layer to use all available
parameters: https://github.com/explosion/thinc/pull/754
This change updates `TextCatBOW` to `v3` which uses the new
`SparseLinear_v2` layer. This results in a sizeable improvement on a
text categorization task that was tested.
While at it, this `spacy.TextCatBOW.v3` also adds the `length_exponent`
option to make it possible to change the hidden size. Ideally, we'd just
have an option called `length`. But the way that `TextCatBOW` uses
hashes results in a non-uniform distribution of parameters when the
length is not a power of two.
* Replace TexCatBOW `length_exponent` parameter by `length`
We now round up the length to the next power of two if it isn't
a power of two.
* Remove some tests for TextCatBOW.v2
* Fix missing import
* add language extensions for norwegian nynorsk and faroese
* update docstring for nn/examples.py
* use relative imports
* add fo and nn tokenizers to pytest fixtures
* add unittests for fo and nn and fix bug in nn
* remove module docstring from fo/__init__.py
* add comments about example sentences' origin
* add license information to faroese data credit
* format unittests using black
* add __init__ files to test/lang/nn and tests/lang/fo
* fix import order and use relative imports in fo/__nit__.py and nn/__init__.py
* Make the tests a bit more compact
* Add fo and nn to website languages
* Add note about jul.
* Add "jul." as exception
---------
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update the "Missing factory" error message
This accounts for model installations that took place during the current Python session.
* Add a note about Jupyter notebooks
* Move error to `spacy.cli.download`
Add extra message for Jupyter sessions
* Add additional note for interactive sessions
* Remove note about `spacy-transformers` from error message
* `isort`
* Improve checks for colab (also helps displacy)
* Update warning messages
* Improve flow for multiple checks
---------
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update Tokenizer.explain for special cases with whitespace
Update `Tokenizer.explain` to skip special case matches if the exact
text has not been matched due to intervening whitespace.
Enable fuzzy `Tokenizer.explain` tests with additional whitespace
normalization.
* Add unit test for special cases with whitespace, xfail fuzzy tests again
* Fix displacy span stacking.
* Format. Remove counter.
* Remove test files.
* Add unit test. Refactor to allow for unit test.
* Fix off-by-one error in tests.
* Add note on score_weight if using a non-default span_key for SpanCat.
* Fix formatting.
* Fix formatting.
* Fix typo.
* Use warning infobox.
* Fix infobox formatting.
tests / Test (macos-latest, 3.11) (push) Has been cancelled
tests / Test (macos-latest, 3.8) (push) Has been cancelled
tests / Test (ubuntu-20.04, 3.6) (push) Has been cancelled
tests / Test (ubuntu-latest, 3.11) (push) Has been cancelled
tests / Test (ubuntu-latest, 3.9) (push) Has been cancelled
tests / Test (windows-latest, 3.10) (push) Has been cancelled
tests / Test (windows-latest, 3.11) (push) Has been cancelled
tests / Test (windows-latest, 3.7) (push) Has been cancelled
* Load the cli module lazily for spacy.info
This avoids that the `spacy` module cannot be imported when the
users chooses not to install `typer`/`requests`.
* Add test
---------
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* adding rolegal model to the spaCy universe
* Fix formatting
* Use raw URL
* update image url and example
* fix pip and update url to raw
* okay, let's add thumb instead of image 🐙
* Update website/meta/universe.json
---------
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* add span key option for CLI evaluation
* Rephrase CLI help to refer to Doc.spans instead of spancat
* Rephrase docs to refer to Doc.spans instead of spancat
---------
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* fix construction example
* shorten task-specific factory list
* small edits to HF models
* small edit to API models
* typo
* fix space
Co-authored-by: Raphael Mitsch <r.mitsch@outlook.com>
---------
Co-authored-by: Raphael Mitsch <r.mitsch@outlook.com>
* initial
* initial documentation run
* fix typo
* Remove mentions of Torchscript and quantization
Both are disabled in the initial release of `spacy-curated-transformers`.
* Fix `piece_encoder` entries
* Remove `spacy-transformers`-specific warning
* Fix duplicate entries in tables
* Doc fixes
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Remove type aliases
* Fix copy-paste typo
* Change `debug pieces` version tag to `3.7`
* Set curated transformers API version to `3.7`
* Fix transformer listener naming
* Add docs for `init fill-config-transformer`
* Update CLI command invocation syntax
* Update intro section of the pipeline component docs
* Fix source URL
* Add a note to the architectures section about the `init fill-config-transformer` CLI command
* Apply suggestions from code review
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update CLI command name, args
* Remove hyphen from the `curated-transformers.mdx` filename
* Fix links
* Remove placeholder text
* Add text to the model/tokenizer loader sections
* Fill in the `DocTransformerOutput` section
* Formatting fixes
* Add curated transformer page to API docs sidebar
* More formatting fixes
* Remove TODO comment
* Remove outdated info about default config
* Apply suggestions from code review
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Add link to HF model hub
* `prettier`
---------
Co-authored-by: Madeesh Kannan <shadeMe@users.noreply.github.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
SpaCy's HashEmbedCNN layer performs convolutions over tokens to produce
contextualized embeddings using a `MaxoutWindowEncoder` layer. These
convolutions are implemented using Thinc's `expand_window` layer, which
concatenates `window_size` neighboring sequence items on either side of
the sequence item being processed. This is repeated across `depth`
convolutional layers.
For example, consider the sequence "ABCDE" and a `MaxoutWindowEncoder`
layer with a context window of 1 and a depth of 2. We'll focus on the
token "C". We can visually represent the contextual embedding produced
for "C" as:
```mermaid
flowchart LR
A0(A<sub>0</sub>)
B0(B<sub>0</sub>)
C0(C<sub>0</sub>)
D0(D<sub>0</sub>)
E0(E<sub>0</sub>)
B1(B<sub>1</sub>)
C1(C<sub>1</sub>)
D1(D<sub>1</sub>)
C2(C<sub>2</sub>)
A0 --> B1
B0 --> B1
C0 --> B1
B0 --> C1
C0 --> C1
D0 --> C1
C0 --> D1
D0 --> D1
E0 --> D1
B1 --> C2
C1 --> C2
D1 --> C2
```
Described in words, this graph shows that before the first layer of the
convolution, the "receptive field" centered at each token consists only
of that same token. That is to say, that we have a receptive field of 1.
The first layer of the convolution adds one neighboring token on either
side to the receptive field. Since this is done on both sides, the
receptive field increases by 2, giving the first layer a receptive field
of 3. The second layer of the convolutions adds an _additional_
neighboring token on either side to the receptive field, giving a final
receptive field of 5.
However, this doesn't match the formula currently given in the docs,
which read:
> The receptive field of the CNN will be
> `depth * (window_size * 2 + 1)`, so a 4-layer network with a window
> size of `2` will be sensitive to 20 words at a time.
Substituting in our depth of 2 and window size of 1, this formula gives
us a receptive field of:
```
depth * (window_size * 2 + 1)
= 2 * (1 * 2 + 1)
= 2 * (2 + 1)
= 2 * 3
= 6
```
This not only doesn't match our computations from above, it's also an
even number! This is suspicious, since the receptive field is supposed
to be centered on a token, and not between tokens. Generally, this
formula results in an even number for any even value of `depth`.
The error in this formula is that the adjustment for the center token
is multiplied by the depth, when it should occur only once. The
corrected formula, `depth * window_size * 2 + 1`, gives the correct
value for our small example from above:
```
depth * window_size * 2 + 1
= 2 * 1 * 2 + 1
= 4 + 1
= 5
```
These changes update the docs to correct the receptive field formula and
the example receptive field size.
There was a mistake in the regex pattern which caused not matching all the desired tokens. The problem was that when we use r string literal prefix to suppose a raw text, we should not use two backslashes to demonstrate a backslash.
* feat: add example stubs
* fix: add required annotations
* fix: mypy issues
* fix: use Py36-compatible Portocol
* Minor reformatting
* adding further type specifications and removing internal methods
* black formatting
* widen type to iterable
* add private methods that are being used by the built-in convertors
* revert changes to corpus.py
* fixes
* fixes
* fix typing of PlainTextCorpus
---------
Co-authored-by: Basile Dura <basile@bdura.me>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Support registered vectors
* Format
* Auto-fill [nlp] on load from config and from bytes/disk
* Only auto-fill [nlp]
* Undo all changes to Language.from_disk
* Expand BaseVectors
These methods are needed in various places for training and vector
similarity.
* isort
* More linting
* Only fill [nlp.vectors]
* Update spacy/vocab.pyx
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Revert changes to test related to auto-filling [nlp]
* Add vectors registry
* Rephrase error about vocab methods for vectors
* Switch to dummy implementation for BaseVectors.to_ops
* Add initial draft of docs
* Remove example from BaseVectors docs
* Apply suggestions from code review
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update website/docs/api/basevectors.mdx
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Fix type and lint bpemb example
* Update website/docs/api/basevectors.mdx
---------
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* remove migration support form
* initial test commit
* add fixture
* add combo test
* pull out parameter example data
* fix formatting on examples
* remove unused import
* remove unncessary fmt:off instructions
* only set logger level if verbose flag is explicitly set
---------
Co-authored-by: svlandeg <svlandeg@github.com>
* Add data structures to docs
* Adjusted descriptions for more consistency
* Add _optional_ flag to parameters
* Add tests and adjust optional title key in doc
* Add title to dep visualizations
* fix typo
---------
Co-authored-by: thomashacker <EdwardSchmuhl@web.de>
* Add cli for finding locations of registered func
* fixes: naming and typing
* isort
* update naming
* remove to find-function
* remove file:// bit
* use registry name if given and exit gracefully if a registry was not found
* clean up failure msg
* specify registry_name options
* mypy fixes
* return location for internal usage
* add documentation
* more mypy fixes
* clean up example
* add section to menu
* add tests
---------
Co-authored-by: svlandeg <svlandeg@github.com>
* Update numpy build constraints for numpy 1.25
Starting in numpy 1.25 (see
https://github.com/numpy/numpy/releases/tag/v1.25.0), the numpy C API is
backwards-compatible by default.
For python 3.9+, we should be able to drop the specific numpy build
requirements and use `numpy>=1.25`, which is currently
backwards-compatible to `numpy>=1.19`.
In the future, the python <3.9 requirements could be dropped and the
lower numpy pin could correspond to the oldest supported version for the
current lower python pin.
* Turn off fail-fast
* Revert "Turn off fail-fast"
This reverts commit 4306f516bc4a6b3437b5393ff1b6b6ae54957d2d.
* Update for python 3.6
* Fix typo
* Update universe.json
* Update universe.json
add some missing commas in the greCy's description.
* Update punctuation.py
Add mathematical left and right angle brackets as punctuation for ancient Greek for better tokenization.
* modified: spacy/language.py
- corrected typo in docstring for :method:`Language.replace_listeners`
- added noqa comment on unused local variable assignment in :method:`Language.from_config` as I wasn't sure if it should be unassigned
modified: website/docs/api/language.mdx
- corrected typo in `Language.replace_listeners` markdown
* modified: spacy/language.py
- removed noqa comment
---------
Co-authored-by: Ian Thompson <ian.thompson@hrblock.com>
These changes add a missing call to `escape_html` in the displaCy span
renderer. Previously span-annotated tokens would be inserted into the
page markup without being escaped, resulting in potentially incorrect
rendering. When I encountered this issue, it resulted in some docs and
span underlines being superimposed on top of properly rendered docs and
span underlines near the beginning of the visualization (due to an
unescaped `<span>` tag).
* Setting up weasel branch (#12456)
* remove project-specific functionality
* remove project-specific tests
* remove project-specific schemas
* remove project-specific information in about
* remove project-specific functions in util.py
* remove project-specific error strings
* remove project-specific CLI commands
* black formatting
* restore some functions that are used beyond projects
* remove project imports
* remove imports
* remove remote_storage tests
* remove one more project unit test
* update for PR 12394
* remove get_hash and get_checksum
* remove upload_ and download_file methods
* remove ensure_pathy
* revert clumsy fingers
* reinstate E970
* feat: use weasel as spacy project command (#12473)
* feat: use weasel as spacy project command
* build: use constrained requirement for weasel
* feat: add weasel to the library requirements
* build: update weasel to new version
* build: use specific weasel tag
* build: use weasel-0.1.0rc1 from PyPI
* fix: remove weasel from requirements.txt
* fix: requirements.txt and setup.cfg need to reflect each other
* feat: remove legacy spacy project code
* bump version
* further merge fixes
* isort
---------
Co-authored-by: Basile Dura <bdura@users.noreply.github.com>
When the default `max_length` is not set and there are longer training
documents, it can be difficult to train and evaluate the span finder due
to memory limits and the time it takes to evaluate a huge number of
predicted spans.
* `Language.replace_listeners`: Pass the replaced listener and the `tok2vec` pipe to the callback
* Update developer docs
* `isort` fixes
* Add error message to assertion
* Add clarification to dev docs
* Replace assertion with exception
* Doc fixes
* Fix problem with universe pages using `docker` language
* Fix problem with universe pages using `r` language
* Add fallback, in case code language is unknown
* Support custom token/lexeme attribute for vectors
* Fix imports
* Back off to ORTH without Vectors.attr
* Fallback if vectors.attr doesn't exist
* Update docs
When sourcing a component, the object from the original pipeline is added to the new pipeline as the same object. This creates a situation where there are several attributes that cannot be in sync between the original pipeline and the new pipeline at the same time for this one object:
* component.name
* component.listener_map / component.listening_components for tok2vec and transformer
When running replace_listeners on a component, the config is not updated correctly if the state of the component is incorrect for the current pipeline (in particular changes that should be applied from model.attrs["replace_listener_cfg"] as used in spacy-transformers) due to the fact that:
* find_listeners relies on component.name to set the name in the listener_map
* replace_listeners relies on listener_map to determine how to modify the configs
In addition, there are several places where pipeline components are modified and the listener map and/or internal component names aren't currently updated.
In cases where there is a component shared by two pipelines that cannot be in sync, this PR chooses to prioritize the most recently modified or initialized pipeline. There is no actual solution with the current source behavior that will make both pipelines usable, so the current pipeline is updated whenever components are added/renamed/removed or the pipeline is initialized for training.
* Add SpanMarker for NER to spaCy universe
* Escape the newlines in the text in the code example
Or at least, attempt to
* Remove now unnecessary import
* Disable NER pipeline component in code example
This is a really odd bug, where Firefox doesn't re-render the `code` element, even though `children` changed.
Two things fixed that:
- remove the `language-ini` `className`
- replace the `code` block with a `div`
Both are not ideal. Therefor this solution adds an inner `div` that now has the classes while still maintaining the semantic `code` element.
I couldn't find any explanation for why this is happening and why it only happens in Firefox. I assume it is a bug caused by one of our many dependencies (or their interplay)
To make matters worse: This bug *doesn't* occure when running the site in dev mode. You have to build and serve the site to recreate it.
* Use isort with Black profile
* isort all the things
* Fix import cycles as a result of import sorting
* Add DOCBIN_ALL_ATTRS type definition
* Add isort to requirements
* Remove isort from build dependencies check
* Typo
* span finder integrated into spacy from experimental
* black
* isort
* black
* default spankey constant
* black
* Update spacy/pipeline/spancat.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* rename
* rename
* max_length and min_length as Optional[int] and strict checking
* black
* mypy fix for integer type infinity
* revert line order
* implement all comparison operators for inf int
* avoid two for loops over all docs by not precomputing
* interleave thresholding with span creation
* black
* revert to not interleaving (relized its faster)
* black
* Update spacy/errors.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* update dosctring
* enforce that the gold and predicted documents have the same text
* new error for ensuring reference and predicted texts are the same
* remove todo
* adjust test
* black
* handle misaligned tokenization
* return correct variable
* failing overfit test
* only use a single spans_key like in spancat
* black
* remove debug lines
* typo
* remove comment
* remove near duplicate reduntant method
* use the 'spans_key' variable name everywhere
* Update spacy/pipeline/span_finder.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* flaky test fix suggestion, hand set bias terms
* only test suggester and test result exhaustively
* make it clear that the span_finder_suggester is more general (not specific to span_finder)
* Update spacy/tests/pipeline/test_span_finder.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Apply suggestions from code review
* remove question comment
* move preset_spans_suggester test to spancat tests
* Add docs and unify default configs for spancat and span finder
* Add `allow_overlap=True` to span finder scorer
* Fix offset bug in set_annotations
* Ignore labels in span finder scorer
* Format
* Add span_finder to quickstart template
* Move settings to self.cfg, store min/max unset as None
* Remove debugging
* Update docstrings and docs
* Update spacy/pipeline/span_finder.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Fix imports
---------
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Require that all SpanGroup spans are from the current doc
The restriction on only adding spans from the current doc were already
implemented for all operations except for `SpanGroup.__init__`.
Initialize copied spans for `SpanGroup.copy` with `Doc.char_span` in
order to validate the character offsets and to make it possible to copy
spans between documents with differing tokenization. Currently there is
no validation that the document texts are identical, but the span char
offsets must be valid spans in the target doc, which prevents you from
ending up with completely invalid spans.
* Undo change in test_beam_overfitting_IO
* Address upcoming numpy v1.25 deprecations in test suite
* Temporarily test most recent numpy prerelease in CI
* Revert "Temporarily test most recent numpy prerelease in CI"
This reverts commit d75a66e55e29e38f8cb3daf4f55a7b87ec13a7fe.
While the typing_extensions/pydantic `Literal` bugs are being sorted
out, disable fail-fast so the rest of the CI is available for
development purposes.
* Add scorer option to return per-component scores
Add `per_component` option to `Language.evaluate` and `Scorer.score` to
return scores keyed by `tokenizer` (hard-coded) or by component name.
Add option to `evaluate` CLI to score by component. Per-component scores
can only be saved to JSON.
* Update help text and messages
This reverts commit 6f314f99c4.
We are reverting this until we can support this normalization more
consistently across vectors, training corpora, and lemmatizer data.
* parsigs universe
* added model installation explanation in the description
* Update website/meta/universe.json
Co-authored-by: Basile Dura <bdura@users.noreply.github.com>
* added model installement instruction in the code example
---------
Co-authored-by: Basile Dura <bdura@users.noreply.github.com>
* Use Latin normalization for Serbian attrs
Use Latin normalization for Serbian `NORM`, `PREFIX`, and `SUFFIX`.
* Update NORMs in tokenizer exceptions and related tests
* Add tests for all custom lex attrs
* Remove unused imports
* Add spans in spacy benchmark
The current implementation of spaCy benchmark accuracy / spacy evaluate
doesn't include the "spans" type, so calling the command doesn't render
the HTML displaCy file needed.
This PR attempts to fix that by creating a new parameter for "spans"
and calling the appropriate displaCy value.
* Reformat file with black
* Add tests for evaluate
* Fix spans -> span for displacy style
* Update test to check render instead
* Update source so mypy passes
* Add parser information to avoid warnings
* CI: Only run test suite once with thinc-apple-ops for macos python 3.11
* Adjust syntax
* Try alternate syntax
* Try alternate syntax
* Try alternate syntax
* avoid nesting then flattening
* mypy fix
* Apply suggestions from code review
* Add type for indices
* Run full matrix for mypy
* Add back modified type: ignore
* Revert "Run full matrix for mypy"
This reverts commit e218873d049d5634e6faa0341ada9af5d53b5a29.
---------
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Add noun chunking to la syntax iterators
* Expand list of numeral, ordinal words
* Expand abbreviations in la tokenizer_exceptions
* Add example sents
* Update spacy/lang/la/syntax_iterators.py
Reorganize la syntax iterators
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Minor updates based on review
* fix call
---------
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Add default to MorphAnalysis.get
Similar to `dict`, allow a `default` option for `MorphAnalysis.get` for
the user to provide a default return value if the field is not found.
The default return value remains `[]`, which is not the same as
`dict.get`, but is already established as this method's default return
value with the return type `List[str]`. However the new `default` option
does not enforce that the user-provided default is actually `List[str]`.
* Restore test case
In `Tokenizer.from_bytes`, the exceptions should be loaded last so that
they are only processed once as part of loading the model.
The exceptions are tokenized as phrase matcher patterns in the
background and the internal tokenization needs to be synced with all the
remaining tokenizer settings. If the exceptions are not loaded last,
there are speed regressions for `Tokenizer.from_bytes/disk` vs.
`Tokenizer.add_special_case` as the caches are reloaded more than
necessary during deserialization.
Replace `progress_bar = "all_steps"` with `progress_bar = "eval"`, which is consistent with the default behavior for `spacy.ConsoleLogger.v1` and `spacy.ConsoleLogger.v2`.
Switch PR tests back to paths-ignore but include changes to `.github`
for all PRs rather than trying to figure out complicated
includes+excludes. Changes to `.github` are relatively rare and should
not be a huge burden for the CI.
* [wip] Update
* [wip] Update
* Add initial port
* [wip] Update
* Fix all imports
* Add spancat_exclusive to pipeline
* [WIP] Update
* [ci skip] Add breakpoint for debugging
* Use spacy.SpanCategorizer.v1 as default archi
* Update spacy/pipeline/spancat_exclusive.py
Co-authored-by: kadarakos <kadar.akos@gmail.com>
* [ci skip] Small updates
* Use Softmax v2 directly from thinc
* Cache the label map
* Fix mypy errors
However, I ignored line 370 because it opened up a bunch of type errors
that might be trickier to solve and might lead to a more complicated
codebase.
* avoid multiplication with 1.0
Co-authored-by: kadarakos <kadar.akos@gmail.com>
* Update spacy/pipeline/spancat_exclusive.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update component versions to v2
* Add scorer to docstring
* Add _n_labels property to SpanCategorizer
Instead of using len(self.labels) in initialize() I am using a private
property self._n_labels. This achieves implementation parity and allows
me to delete the whole initialize() method for spancat_exclusive (since
it's now the same with spancat).
* Inherit from SpanCat instead of TrainablePipe
This commit changes the inheritance structure of Exclusive_Spancat,
now it's inheriting from SpanCategorizer than TrainablePipe. This
allows me to remove duplicate methods that are already present in
the parent function.
* Revert documentation link to spancat
* Fix init call for exclusive spancat
* Update spacy/pipeline/spancat_exclusive.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Import Suggester from spancat
* Include zero_init.v1 for spancat
* Implement _allow_extra_label to use _n_labels
To ensure that spancat / spancat_exclusive cannot be resized after
initialization, I inherited the _allow_extra_label() method from
spacy/pipeline/trainable_pipe.pyx and used self._n_labels instead
of len(self.labels) for checking.
I think that changing it locally is a better solution rather than
forcing each class that inherits TrainablePipe to use the self._n_labels
attribute.
Also note that I turned-off black formatting in this block of code
because it reads better without the overhang.
* Extend existing tests to spancat_exclusive
In this commit, I extended the existing tests for spancat to include
spancat_exclusive. I parametrized the test functions with 'name'
(similar var name with textcat and textcat_multilabel) for each
applicable test.
TODO: Add overfitting tests for spancat_exclusive
* Update documentation for spancat
* Turn on formatting for allow_extra_label
* Remove initializers in default config
* Use DEFAULT_EXCL_SPANCAT_MODEL
I also renamed spancat_exclusive_default_config into
spancat_excl_default_config because black does some not pretty
formatting changes.
* Update documentation
Update grammar and usage
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Clarify docstring for Exclusive_SpanCategorizer
* Remove mypy ignore and typecast labels to list
* Fix documentation API
* Use a single variable for tests
* Update defaults for number of rows
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Put back initializers in spancat config
Whenever I remove model.scorer.init_w and model.scorer.init_b,
I encounter an error in the test:
SystemError: <method '__getitem__' of 'dict' objects> returned a result
with an error set.
My Thinc version is 8.1.5, but I can't seem to check what's causing the
error.
* Update spancat_exclusive docstring
* Remove init_W and init_B parameters
This commit is expected to fail until the new Thinc release.
* Require thinc>=8.1.6 for serializable Softmax defaults
* Handle zero suggestions to make tests pass
I'm not sure if this is the most elegant solution. But what should
happen is that the _make_span_group function MUST return an empty
SpanGroup if there are no suggestions.
The error happens when the 'scores' variable is empty. We cannot
get the 'predicted' and other downstream vars.
* Better approach for handling zero suggestions
* Update website/docs/api/spancategorizer.md
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update spancategorizer headers
* Apply suggestions from code review
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Add default value in negative_weight in docs
* Add default value in allow_overlap in docs
* Update how spancat_exclusive is constructed
In this commit, I added the following:
- Put the default values of negative_weight and allow_overlap
in the default_config dictionary.
- Rename make_spancat -> make_exclusive_spancat
* Run prettier on spancategorizer.mdx
* Change exactly one -> at most one
* Add suggester documentation in Exclusive_SpanCategorizer
* Add suggester to spancat docstrings
* merge multilabel and singlelabel spancat
* rename spancat_exclusive to singlelable
* wire up different make_spangroups for single and multilabel
* black
* black
* add docstrings
* more docstring and fix negative_label
* don't rely on default arguments
* black
* remove spancat exclusive
* replace single_label with add_negative_label and adjust inference
* mypy
* logical bug in configuration check
* add spans.attrs[scores]
* single label make_spangroup test
* bugfix
* black
* tests for make_span_group with negative labels
* refactor make_span_group
* black
* Update spacy/tests/pipeline/test_spancat.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* remove duplicate declaration
* Update spacy/pipeline/spancat.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* raise error instead of just print
* make label mapper private
* update docs
* run prettier
* Update website/docs/api/spancategorizer.mdx
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update website/docs/api/spancategorizer.mdx
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update spacy/pipeline/spancat.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update spacy/pipeline/spancat.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update spacy/pipeline/spancat.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update spacy/pipeline/spancat.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* don't keep recomputing self._label_map for each span
* typo in docs
* Intervals to private and document 'name' param
* Update spacy/pipeline/spancat.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update spacy/pipeline/spancat.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* add Tag to new features
* replace tags
* revert
* revert
* revert
* revert
* Update website/docs/api/spancategorizer.mdx
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update website/docs/api/spancategorizer.mdx
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* prettier
* Fix merge
* Update website/docs/api/spancategorizer.mdx
* remove references to 'single_label'
* remove old paragraph
* Add spancat_singlelabel to config template
* Format
* Extend init config tests
---------
Co-authored-by: kadarakos <kadar.akos@gmail.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Change GPU efficient textcat to use CNN, not BOW
If you generate a config with a textcat component using GPU
(transformers), the defaut option (efficiency) uses a BOW architecture,
which does not use tok2vec features. While that can make sense as part
of a larger pipeline, in the case of just a transformer and a textcat,
that means the transformer is doing a lot of work for no purpose.
This changes it so that the CNN architecture is used instead. It could
also be changed to be the same as the accuracy config, which uses the
ensemble architecture.
* Add the transformer when using a textcat with GPU
* Switch ubuntu-latest to ubuntu-20.04 in main tests (#11928)
* Switch ubuntu-latest to ubuntu-20.04 in main tests
* Only use 20.04 for 3.6
* Require thinc v8.1.7
* Require thinc v8.1.8
* Break up longer expression
---------
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Handle deprecation of pkg_resources
* Replace `pkg_resources` with `importlib_metadata` for `spacy info
--url`
* Remove requirements check from `spacy project` given the lack of
alternatives
* Fix installed model URL method and CI test
* Fix types/handling, simplify catch-all return
* Move imports instead of disabling requirements check
* Format
* Reenable test with ignored deprecation warning
* Fix except
* Fix return
* Make empty_kb() configurable.
* Format.
* Update docs.
* Be more specific in KB serialization test.
* Update KB serialization tests. Update docs.
* Remove doc update for batched candidate generation.
* Fix serialization of subclassed KB in tests.
* Format.
* Update docstring.
* Update docstring.
* Switch from pickle to json for custom field serialization.
* Add immediate left/right child/parent dependency relations
* Add tests for new REL_OPs: `>+`, `>-`, `<+`, and `<-`.
---------
Co-authored-by: Tan Long <tanloong@foxmail.com>
* add unittest for explosion#12311
* create punctuation.py for swedish
* removed : from infixes in swedish punctuation.py
* allow : as infix if succeeding char is uppercase
* standardize predicate key format
* single key function
* Make optional args in key function keyword-only
---------
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Fix FUZZY operator definition
The default length of the FUZZY operator is 2 and not 3.
* adjust edit distance in matcher usage docs too
---------
Co-authored-by: svlandeg <svlandeg@github.com>
* change logging call for spacy.LookupsDataLoader.v1
* substitutions in language and _util
* various more substitutions
* add string formatting guidelines to contribution guidelines
* Normalize whitespace in evaluate CLI output test
Depending on terminal settings, lines may be padded to the screen width
so the comparison is too strict with only the command string replacement.
* Move to test util method
* Change to normalization method
* Normalize whitespace in evaluate CLI output test
Depending on terminal settings, lines may be padded to the screen width
so the comparison is too strict with only the command string replacement.
* Move to test util method
* Change to normalization method
* Add span_id to Span.char_span, update Doc/Span.char_span docs
`Span.char_span(id=)` should be removed in the future.
* Also use Union[int, str] in Doc docstring
* WIP
* rm ipython embeds
* rm total
* WIP
* cleanup
* cleanup + reword
* rm component function
* remove migration support form
* fix reference dataset for dev data
* additional fixes
- set approach to identifying unique trees
- adjust line length on messages
- add logic for detecting docs without annotations
* use 0 instead of none for no annotation
* partial annotation support
* initial tests for _compile_gold lemma attributes
Using the example data from the edit tree lemmatizer tests for:
- lemmatizer_trees
- partial_lemma_annotations
- n_low_cardinality_lemmas
- no_lemma_annotations
* adds output test for cli app
* switch msg level
* rm unclear uniqueness check
* Revert "rm unclear uniqueness check"
This reverts commit 6ea2b3524bd66417b3d54854bb7d27ea7e6029ae.
* remove good message on uniqueness
* formatting
* use en_vocab fixture
* clarify data set source in messages
* remove unnecessary import
Co-authored-by: svlandeg <svlandeg@github.com>
* Add `spacy.PlainTextCorpusReader.v1`
This is a corpus reader that reads plain text corpora with the following
format:
- UTF-8 encoding
- One line per document.
- Blank lines are ignored.
It is useful for applications where we deal with very large corpora,
such as distillation, and don't want to deal with the space overhead of
serialized formats. Additionally, many large corpora already use such
a text format, keeping the necessary preprocessing to a minimum.
* Update spacy/training/corpus.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* docs: add version to `PlainTextCorpus`
* Add docstring to registry function
* Add plain text corpus tests
* Only strip newline/carriage return
* Add return type _string_to_tmp_file helper
* Use a temporary directory in place of file name
Different OS auto delete/sharing semantics are just wonky.
* This will be new in 3.5.1 (rather than 4)
* Test improvements from code review
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Rename CSS class to make use more clear
* Rename component prop to improve code readability
* Fix `aria-hidden` directly on a link element
This link wouldn't have been clickable by screenreaders
* Refactor component
This removes a unnessary `div` and a duplicate link
Co-authored-by: Ines Montani <ines@ines.io>
Originally introduced in 62b9c9c6d711584cea91244bc765e56e74337fdd
Original error: Warning: Invalid DOM property `class`. Did you mean `className`?
React doesn't have `class`, it uses `className`.
* Fix missing comma
* Activate user zoom for website
This is recommended by lighthouse:
> Disabling zooming is problematic for users with low vision who rely on screen magnification to properly see the contents of a web page. Learn more.
Also iOS already ignores this attribute anyway.
* Fix gap in landing pattern at the top
* Replace `.jpg` patterns with `.png`
This drastically reduces file size (for the landing page from 221kb to 57kb) while doubling the resolution to look sharper on retina displays.
* Refactor _scores2guesses
* Handle arrays on GPU
* Convert argmax result to raw integer
Co-authored-by: Madeesh Kannan <shadeMe@users.noreply.github.com>
* Use NumpyOps() to copy data to CPU
Co-authored-by: Madeesh Kannan <shadeMe@users.noreply.github.com>
* Changes based on review comments
* Use different _scores2guesses depending on tree_k
* Add tests for corner cases
* Add empty line for consistency
* Improve naming
Co-authored-by: Daniël de Kok <me@github.danieldk.eu>
* Improve naming
Co-authored-by: Daniël de Kok <me@github.danieldk.eu>
Co-authored-by: Madeesh Kannan <shadeMe@users.noreply.github.com>
Co-authored-by: Daniël de Kok <me@github.danieldk.eu>
* skeleton
* Fill in non-CLI details from release notes draft
* Add TODO for fuzzy matching
* Website updates for v3-5 draft
* Fill in usage examples
* Add fuzzy matching to intro
* Fix fuzzy examples
* Shell example formatting
* Fix typo
* Format
* Remove trailing periods in internal list
* Update
* Fix spacing for nested lists
* Update InMemoryLookupKB link
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Co-authored-by: Ines Montani <ines@ines.io>
* API docs: Rename kb_in_memory to inmemorylookupkb, add to sidebar
* adjust to mdx
* linkout to InMemoryLookupKB at first occurrence in kb.mdx
* fix links to docs
* revert Azure trigger setting (I'll make a separate PR)
Co-authored-by: svlandeg <svlandeg@github.com>
* Update Dockerfile to work with Next.js
- Update to Node 18
- Do not run as root, this also works better with Node
privilege-dropping.
- Update README with new run instructions and adding the
`--rm` flag to avoid leaving a bunch of unused Docker
containers.
- Also change README to recommend building the image locally.
Image builds are pretty fast and the uploaded images get
outdated pretty quickly.
* Add .dockerignore to avoid sending large build contexts
* Typo
* Clarify how `--code` arg works
This adds a few sentences to the docs to clarify how the `--code`
argument works, including an explanation of how to load custom
components in your own code.
* Add link to spacy.load docs
* Add a `spacy evaluate speed` subcommand
This subcommand reports the mean batch performance of a model on a data set with
a 95% confidence interval. For reliability, it first performs some warmup
rounds. Then it will measure performance on batches with randomly shuffled
documents.
To avoid having too many spaCy commands, `speed` is a subcommand of `evaluate`
and accuracy evaluation is moved to its own `evaluate accuracy` subcommand.
* Fix import cycle
* Restore `spacy evaluate`, make `spacy benchmark speed` an alias
* Add documentation for `spacy benchmark`
* CREATES -> PRINTS
* WPS -> words/s
* Disable formatting of benchmark speed arguments
* Fail with an error message when trying to speed bench empty corpus
* Make it clearer that `benchmark accuracy` is a replacement for `evaluate`
* Fix docstring webpage reference
* tests: check `evaluate` output against `benchmark accuracy`
* Clean up displacy port-related error messages, docs
There were some issues in the error messages and docs in #11948.
1. the error messages didn't specify the port argument to displacy.serve correctly
2. the docs didn't mark the auto select argument as new
This addresses those issues.
* Update website/docs/api/top-level.md
Co-authored-by: Raphael Mitsch <r.mitsch@outlook.com>
* Apply prettier
Co-authored-by: Raphael Mitsch <r.mitsch@outlook.com>
* Rename all MDX file to `.mdx`
* Lock current node version (#11885)
* Apply Prettier (#11996)
* Minor website fixes (#11974) [ci skip]
* fix table
* Migrate to Next WEB-17 (#12005)
* Initial commit
* Run `npx create-next-app@13 next-blog`
* Install MDX packages
Following: https://github.com/vercel/next.js/blob/77b5f79a4dff453abb62346bf75b14d859539b81/packages/next-mdx/readme.md
* Add MDX to Next
* Allow Next to handle `.md` and `.mdx` files.
* Add VSCode extension recommendation
* Disabled TypeScript strict mode for now
* Add prettier
* Apply Prettier to all files
* Make sure to use correct Node version
* Add basic implementation for `MDXRemote`
* Add experimental Rust MDX parser
* Add `/public`
* Add SASS support
* Remove default pages and styling
* Convert to module
This allows to use `import/export` syntax
* Add import for custom components
* Add ability to load plugins
* Extract function
This will make the next commit easier to read
* Allow to handle directories for page creation
* Refactoring
* Allow to parse subfolders for pages
* Extract logic
* Redirect `index.mdx` to parent directory
* Disabled ESLint during builds
* Disabled typescript during build
* Remove Gatsby from `README.md`
* Rephrase Docker part of `README.md`
* Update project structure in `README.md`
* Move and rename plugins
* Update plugin for wrapping sections
* Add dependencies for plugin
* Use plugin
* Rename wrapper type
* Simplify unnessary adding of id to sections
The slugified section ids are useless, because they can not be referenced anywhere anyway. The navigation only works if the section has the same id as the heading.
* Add plugin for custom attributes on Markdown elements
* Add plugin to readd support for tables
* Add plugin to fix problem with wrapped images
For more details see this issue: https://github.com/mdx-js/mdx/issues/1798
* Add necessary meta data to pages
* Install necessary dependencies
* Remove outdated MDX handling
* Remove reliance on `InlineList`
* Use existing Remark components
* Remove unallowed heading
Before `h1` components where not overwritten and would never have worked and they aren't used anywhere either.
* Add missing components to MDX
* Add correct styling
* Fix broken list
* Fix broken CSS classes
* Implement layout
* Fix links
* Fix broken images
* Fix pattern image
* Fix heading attributes
* Rename heading attribute
`new` was causing some weird issue, so renaming it to `version`
* Update comment syntax in MDX
* Merge imports
* Fix markdown rendering inside components
* Add model pages
* Simplify anchors
* Fix default value for theme
* Add Universe index page
* Add Universe categories
* Add Universe projects
* Fix Next problem with copy
Next complains when the server renders something different then the client, therfor we move the differing logic to `useEffect`
* Fix improper component nesting
Next doesn't allow block elements inside a `<p>`
* Replace landing page MDX with page component
* Remove inlined iframe content
* Remove ability to inline HTML content in iFrames
* Remove MDX imports
* Fix problem with image inside link in MDX
* Escape character for MDX
* Fix unescaped characters in MDX
* Fix headings with logo
* Allow to export static HTML pages
* Add prebuild script
This command is automatically run by Next
* Replace `svg-loader` with `react-inlinesvg`
`svg-loader` is no longer maintained
* Fix ESLint `react-hooks/exhaustive-deps`
* Fix dropdowns
* Change code language from `cli` to `bash`
* Remove unnessary language `none`
* Fix invalid code language
`markdown_` with an underscore was used to basically turn of syntax highlighting, but using unknown languages know throws an error.
* Enable code blocks plugin
* Readd `InlineCode` component
MDX2 removed the `inlineCode` component
> The special component name `inlineCode` was removed, we recommend to use `pre` for the block version of code, and code for both the block and inline versions
Source: https://mdxjs.com/migrating/v2/#update-mdx-content
* Remove unused code
* Extract function to own file
* Fix code syntax highlighting
* Update syntax for code block meta data
* Remove unused prop
* Fix internal link recognition
There is a problem with regex between Node and browser, and since Next runs the component on both, this create an error.
`Prop `rel` did not match. Server: "null" Client: "noopener nofollow noreferrer"`
This simplifies the implementation and fixes the above error.
* Replace `react-helmet` with `next/head`
* Fix `className` problem for JSX component
* Fix broken bold markdown
* Convert file to `.mjs` to be used by Node process
* Add plugin to replace strings
* Fix custom table row styling
* Fix problem with `span` inside inline `code`
React doesn't allow a `span` inside an inline `code` element and throws an error in dev mode.
* Add `_document` to be able to customize `<html>` and `<body>`
* Add `lang="en"`
* Store Netlify settings in file
This way we don't need to update via Netlify UI, which can be tricky if changing build settings.
* Add sitemap
* Add Smartypants
* Add PWA support
* Add `manifest.webmanifest`
* Fix bug with anchor links after reloading
There was no need for the previous implementation, since the browser handles this nativly. Additional the manual scrolling into view was actually broken, because the heading would disappear behind the menu bar.
* Rename custom event
I was googeling for ages to find out what kind of event `inview` is, only to figure out it was a custom event with a name that sounds pretty much like a native one. 🫠
* Fix missing comment syntax highlighting
* Refactor Quickstart component
The previous implementation was hidding the irrelevant lines via data-props and dynamically generated CSS. This created problems with Next and was also hard to follow. CSS was used to do what React is supposed to handle.
The new implementation simplfy filters the list of children (React elements) via their props.
* Fix syntax highlighting for Training Quickstart
* Unify code rendering
* Improve error logging in Juniper
* Fix Juniper component
* Automatically generate "Read Next" link
* Add Plausible
* Use recent DocSearch component and adjust styling
* Fix images
* Turn of image optimization
> Image Optimization using Next.js' default loader is not compatible with `next export`.
We currently deploy to Netlify via `next export`
* Dont build pages starting with `_`
* Remove unused files
* Add Next plugin to Netlify
* Fix button layout
MDX automatically adds `p` tags around text on a new line and Prettier wants to put the text on a new line. Hacking with JSX string.
* Add 404 page
* Apply Prettier
* Update Prettier for `package.json`
Next sometimes wants to patch `package-lock.json`. The old Prettier setting indended with 4 spaces, but Next always indends with 2 spaces. Since `npm install` automatically uses the indendation from `package.json` for `package-lock.json` and to avoid the format switching back and forth, both files are now set to 2 spaces.
* Apply Next patch to `package-lock.json`
When starting the dev server Next would warn `warn - Found lockfile missing swc dependencies, patching...` and update the `package-lock.json`. These are the patched changes.
* fix link
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* small backslash fixes
* adjust to new style
Co-authored-by: Marcus Blättermann <marcus@essenmitsosse.de>
In the v3 scorer refactoring, `token_acc` was implemented incorrectly.
It should use `precision` instead of `fscore` for the measure of
correctly aligned tokens / number of predicted tokens.
Fix the docs to reflect that the measure uses the number of predicted
tokens rather than the number of gold tokens.
* enable fuzzy matching
* add fuzzy param to EntityMatcher
* include rapidfuzz_capi
not yet used
* fix type
* add FUZZY predicate
* add fuzzy attribute list
* fix type properly
* tidying
* remove unnecessary dependency
* handle fuzzy sets
* simplify fuzzy sets
* case fix
* switch to FUZZYn predicates
use Levenshtein distance.
remove fuzzy param.
remove rapidfuzz_capi.
* revert changes added for fuzzy param
* switch to polyleven
(Python package)
* enable fuzzy matching
* add fuzzy param to EntityMatcher
* include rapidfuzz_capi
not yet used
* fix type
* add FUZZY predicate
* add fuzzy attribute list
* fix type properly
* tidying
* remove unnecessary dependency
* handle fuzzy sets
* simplify fuzzy sets
* case fix
* switch to FUZZYn predicates
use Levenshtein distance.
remove fuzzy param.
remove rapidfuzz_capi.
* revert changes added for fuzzy param
* switch to polyleven
(Python package)
* fuzzy match only on oov tokens
* remove polyleven
* exclude whitespace tokens
* don't allow more edits than characters
* fix min distance
* reinstate FUZZY operator
with length-based distance function
* handle sets inside regex operator
* remove is_oov check
* attempt build fix
no mypy failure locally
* re-attempt build fix
* don't overwrite fuzzy param value
* move fuzzy_match
to its own Python module to allow patching
* move fuzzy_match back inside Matcher
simplify logic and add tests
* Format tests
* Parametrize fuzzyn tests
* Parametrize and merge fuzzy+set tests
* Format
* Move fuzzy_match to a standalone method
* Change regex kwarg type to bool
* Add types for fuzzy_match
- Refactor variable names
- Add test for symmetrical behavior
* Parametrize fuzzyn+set tests
* Minor refactoring for fuzz/fuzzy
* Make fuzzy_match a Matcher kwarg
* Update type for _default_fuzzy_match
* don't overwrite function param
* Rename to fuzzy_compare
* Update fuzzy_compare default argument declarations
* allow fuzzy_compare override from EntityRuler
* define new Matcher keyword arg
* fix type definition
* Implement fuzzy_compare config option for EntityRuler and SpanRuler
* Rename _default_fuzzy_compare to fuzzy_compare, remove from reexported objects
* Use simpler fuzzy_compare algorithm
* Update types
* Increase minimum to 2 in fuzzy_compare to allow one transposition
* Fix predicate keys and matching for SetPredicate with FUZZY and REGEX
* Add FUZZY6..9
* Add initial docs
* Increase default fuzzy to rounded 30% of pattern length
* Update docs for fuzzy_compare in components
* Update EntityRuler and SpanRuler API docs
* Rename EntityRuler and SpanRuler setting to matcher_fuzzy_compare
To having naming similar to `phrase_matcher_attr`, rename
`fuzzy_compare` setting for `EntityRuler` and `SpanRuler` to
`matcher_fuzzy_compare. Organize next to `phrase_matcher_attr` in docs.
* Fix schema aliases
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Fix typo
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Add FUZZY6-9 operators and update tests
* Parameterize test over greedy
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Fix type for fuzzy_compare to remove Optional
* Rename to spacy.levenshtein_compare.v1, move to spacy.matcher.levenshtein
* Update docs following levenshtein_compare renaming
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* check port in use and add itself
* check port in use and add itself
* Auto switch to nearest available port.
* Use bind to check port instead of connect_ex.
* Reformat.
* Add auto_select_port argument.
* update docs for displacy.serve
* Update spacy/errors.py
Co-authored-by: Paul O'Leary McCann <polm@dampfkraft.com>
* Update website/docs/api/top-level.md
Co-authored-by: Paul O'Leary McCann <polm@dampfkraft.com>
* Update spacy/errors.py
Co-authored-by: Paul O'Leary McCann <polm@dampfkraft.com>
* Add test using multiprocessing
* fix argument name
* Increase sleep times
Want to rule this out as a cause of test failure
* Don't terminate a process that isn't alive
* Refactor port finding logic
This moves all the port logic into its own util function, which can be
tested without having to background a server directly.
* Use with for the server
This ensures the server is closed correctly.
* Pass in the host when checking port availability
* Shorten argument name
* Update error codes following merge
* Add types for arguments, specify docstrings.
* Add typing for arguments with default value.
* Update docstring to match spaCy format.
* Update docstring to match spaCy format.
* Fix docs
Arg name changed from `auto_select_port` to just `auto_select`.
* Revert "Fix docs"
This reverts commit 356966fe849660c0c08b670c6aee1aa2af05c1c1.
Co-authored-by: zhiiw <1302593554@qq.com>
Co-authored-by: Paul O'Leary McCann <polm@dampfkraft.com>
Co-authored-by: Raphael Mitsch <r.mitsch@outlook.com>
* add test for running evaluate on an nlp pipeline with two distinct textcat components
* cleanup
* merge dicts instead of overwrite
* don't add more labels to the given set
* Revert "merge dicts instead of overwrite"
This reverts commit 89bee0ed7798389e6de882a0234e6075fbdaf331.
* Switch tests to separate scorer keys rather than merged dicts
* Revert unrelated edits
* Switch textcat scorers to v2
* formatting
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* fix processing of "auto" in walk_directory
* add check for None
* move AUTO check to convert and fix verification of args
* add specific CLI test with CliRunner
* cleanup
* more cleanup
* update docstring
* Fix inconsistency in displaCy docs about page option
The `page` option, which wraps the output SVG in HTML, is true by
default for `serve` but not for `render`. The `render` docs were wrong
though, so this updates them.
* Update the same statement in more docs
A few renderers used the same language
* Add spacy-pythainlp
* Move submission to right section
* Minor cleanup
* Remove extra list call
* Update universe.json
Co-authored-by: Paul O'Leary McCann <polm@dampfkraft.com>
* Add `ConsoleLogger.v3`
This addition expands the progress bar feature to count up the training/distillation steps to either the next evaluation pass or the maximum number of steps.
* Rename progress bar types
* Add defaults to docs
Minor fixes
* Move comment
* Minor punctuation fixes
* Explicitly check for `None` when validating progress bar type
Co-authored-by: Paul O'Leary McCann <polm@dampfkraft.com>
* Convert all individual values explicitly to uint64 for array-based doc representations
* Temporarily test with latest numpy v1.24.0rc
* Remove unnecessary conversion from attr_t
* Reduce number of individual casts
* Convert specifically from int32 to uint64
* Revert "Temporarily test with latest numpy v1.24.0rc"
This reverts commit eb0e3c5006515b9a7ff52bae59484c909b8a3f65.
* Also use int32 in tests
* Extend to wasabi v1.1
* Temporarily run mypy and tests with newest wasabi
* Temporarily skip check requirements test
* Revert "Temporarily skip check requirements test"
This reverts commit 44f4ce20a8e8c92e8bfc8042cc68333589a96253.
* Revert "Temporarily run mypy and tests with newest wasabi"
This reverts commit e677a2257ced55e696cafc3a8e89eb2f7ddfc369.
Strings in replacement nodes where not added to the `StringStore`
when `EditTreeLemmatizer` was initialized from a set of labels. The
corresponding test did not capture this because it added the strings
through the examples that were passed to the initialization.
This change fixes both this bug in the initialization as the 'shadowing'
of the bug in the test.
If you don't have spacy-transformers installed, but try to use `init
config` with the GPU flag, you'll get an error. The issue is that the
`use_transformers` flag in the config is conflated with the GPU flag,
and then there's an attempt to access transformers config info that may
not exist.
There may be a better way to do this, but this stops the error.
* Support local filesystem remotes for projects
* Fix support for local filesystem remotes for projects
* Use `FluidPath` instead of `Pathy` to support both filesystem and
remote paths
* Create missing parent directories if required for local filesystem
* Add a more general `_file_exists` method to support both `Pathy`,
`Path`, and `smart_open`-compatible URLs
* Add explicit `smart_open` dependency starting with support for
`compression` flag
* Update `pathy` dependency to exclude older versions that aren't
compatible with required `smart_open` version
* Update docs to refer to `Pathy` instead of `smart_open` for project
remotes (technically you can still push to any `smart_open`-compatible
path but you can't pull from them)
* Add tests for local filesystem remotes
* Update pathy for general BlobStat sorting
* Add import
* Remove _file_exists since only Pathy remotes are supported
* Format CLI docs
* Clean up merge
* pymorph2 issues #11620, #11626, #11625:
- #11620: pymorphy2_lookup
- #11626: handle multiple forms pointing to the same normal form + handling empty POS tag
- #11625: matching DET that are labelled as PRON by pymorhp2
* Move lemmatizer algorithm changes back into RussianLemmatizer
* Fix uk pymorphy3_lookup mode init
* Move and update tests for ru/uk lookup lemmatizer modes
* Fix typo
* Remove traces of previous behavior for uninflected POS
* Refactor to private generic-looking pymorphy methods
* Remove xfailed uk lemmatizer cases
* Update spacy/lang/ru/lemmatizer.py
Co-authored-by: Richard Hudson <richard@explosion.ai>
Co-authored-by: Dmytro S Lituiev <d.lituiev@gmail.com>
Co-authored-by: Richard Hudson <richard@explosion.ai>
* Add `training.before_update` callback
This callback can be used to implement training paradigms like gradual (un)freezing of components (e.g: the Transformer) after a certain number of training steps to mitigate catastrophic forgetting during fine-tuning.
* Fix type annotation, default config value
* Generalize arguments passed to the callback
* Update schema
* Pass `epoch` to callback, rename `current_step` to `step`
* Add test
* Simplify test
* Replace config string with `spacy.blank`
* Apply suggestions from code review
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Cleanup imports
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Check textcat values for validity
* Fix error numbers
* Clean up vals reference
* Check category value validity through training
The _validate_categories is called in update, which for multilabel is
inherited from the single label component.
* Formatting
* Add equality definition for vectors
This re-uses the check from sourcing components.
* Use the equality check
* Format
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Fix typos, add couple of new abbreviations, remove nonbreaking spaces
* Remove space from abbreviation
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update warning, add tests for project requirements check
* Make warning more general for differences between PEP 508 and pip
* Add tests for _check_requirements
* Parameterize test
* Add fallback in requirements check, only check once
* Rename to skip_requirements_check
* Update spacy/cli/project/run.py
Co-authored-by: Paul O'Leary McCann <polm@dampfkraft.com>
Co-authored-by: Paul O'Leary McCann <polm@dampfkraft.com>
* Fix default parameters for load functions
Some load functions used SimpleFrozenList() directly instead of the
_DEFAULT_EMPTY_PIPES parameter. That mostly worked as intended, but
the changes in #11459 check for equality using identity, not value, so a
warning is incorrectly raised sometimes, as in #11706.
This change just has all the load functions use the singleton value
instead.
* Add test that there are no warnings on module-based load
This will succeed due to changes in this branch, but local tests with
the latest release failed as intended.
* Try reverting commit and see if CI changes
There is an error in CI that is probably unrelated.
Revert "Fix default parameters for load functions"
This reverts commit dc46b35687e92e4793e64edb11997d44b88c6a8b.
* Revert "Try reverting commit and see if CI changes"
This reverts commit 2514ed07ef29851b5ac60015442a7ce44c69decc.
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Use `build` instead of `python setup.py sdist`
* Remove in-place build with `setup.py`
* Remove `gpu` parameter and GPU tests
* Keep `architecture` and `num_build_jobs` in azure steps with CI
defaults
* Fix use of `num_build_jobs` parameters
* Remove now-unused `prefix` parameter
* Test imports and CLI before installing test requirements
* Remove `*.egg-info` directory in addition to source directory for an
warning-free `import spacy`
* Switch `thinc-apple-ops` test to python 3.11 (as most recent python
that is tested across platforms)
* Update textcat scorer threshold behavior
For `textcat` (with exclusive classes) the scorer should always use a
threshold of 0.0 because there should be one predicted label per doc and
the numeric score for that particular label should not matter.
* Rename to test_textcat_multilabel_threshold
* Remove all uses of threshold for multi_label=False
* Update Scorer.score_cats API docs
* Add tests for score_cats with thresholds
* Update textcat API docs
* Fix types
* Convert threshold back to float
* Fix threshold type in docstring
* Improve formatting in Scorer API docs
* Handle docs with no entities
If a whole batch contains no entities it won't make it to the model, but
it's possible for individual Docs to have no entities. Before this
commit, those Docs would cause an error when attempting to concatenate
arrays because the dimensions didn't match.
It turns out the process of preparing the Ragged at the end of the span
maker forward was a little different from list2ragged, which just uses
the flatten function directly. Letting list2ragged do the conversion
avoids the dimension issue.
This did not come up before because in NEL demo projects it's typical
for data with no entities to be discarded before it reaches the NEL
component.
This includes a simple direct test that shows the issue and checks it's
resolved. It doesn't check if there are any downstream changes, so a
more complete test could be added. A full run was tested by adding an
example with no entities to the Emerson sample project.
* Add a blank instance to default training data in tests
Rather than adding a specific test, since not failing on instances with
no entities is basic functionality, it makes sense to add it to the
default set.
* Fix without modifying architecture
If the architecture is modified this would have to be a new version, but
this change isn't big enough to merit that.
* added spacy-cleaner to the spaCy universe
* Move data to righ section of universe.json
* Cleanup
- fix typo ("replacers")
- spaCy doesn't need to be marked as code
- lemma of "Hello" is lower case
Co-authored-by: Paul O'Leary McCann <polm@dampfkraft.com>
* Fix multiple extensions and character offset
* Rename token_start/end to start/end
* Refactor Doc.from_json based on review
* Iterate over user_data items
* Only add non-empty underscore entries
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Fix flag handling in dvc
Prior to this commit, if a flag (--verbose or --quiet) was passed to
DVC, it would be added to the end of the generated dvc command line.
This would result in the command being interpreted as part of the actual
command to run, rather than an argument to dvc. This would result in
command lines like:
spacy project run preprocess --verbose
That would fail with an error that there's no such directory as
`--verbose`.
This change puts the flags at the front of the dvc command so that they
are interpreted correctly. It removes the `run_dvc_commands` function,
which had been reduced to just a for loop and wasn't used elsewhere.
A separate problem is that there's no way to specify the quiet behaviour
to dvc from the command line, though it's unclear if that's a bug.
* Add dvc quiet flag to docs
* Handle case in DVC where no commands are appropriate
If only have commands with no deps or outputs (admittedly unlikely), you
get a weird error about the dvc file not existing. This gives explicit
output instead.
* Add support for quiet flag
* Fix command execution
Commands are strings now because they're joined further up.
* Fix example code for spacy-wordnet
It looks like in the most recent version, 0.1.0, it's no longer possible
to pass the lang parameter to the component separately. Doing so will
raise an error.
* Apply suggestions from code review
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Cleanup
* More cleanup
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* empty commit
* restrict importlib-metadata to lower than 5.0.0
* restrict importlib-metadata also for validate CI step
* set fixed version for CI
* try flake8 5.0.4 in CI validation step
* from importlib-metadata from requirements again
* Change enable/disable behavior so that arguments take precedence over config options. Extend error message on conflict. Add warning message in case of overwriting config option with arguments.
* Fix tests in test_serialize_pipeline.py to reflect changes to handling of enable/disable.
* Fix type issue.
* Move comment.
* Move comment.
* Issue UserWarning instead of printing wasabi message. Adjust test.
* Added pytest.warns(UserWarning) for expected warning to fix tests.
* Update warning message.
* Move type handling out of fetch_pipes_status().
* Add global variable for default value. Use id() to determine whether used values are default value.
* Fix default value for disable.
* Rename DEFAULT_PIPE_STATUS to _DEFAULT_EMPTY_PIPES.
* add punctuation to grc
Add support for special editorial punctuation that is common in ancient Greek texts. Ancient Greek texts, as found in digital and print form, have been largely edited by scholars. Restorations and improvements are normally marked with special characters that need to be handled properly by the tokenizer.
* add unit tests
* simplify regex
* move generic quotes to char classes
* rename unit test
* fix regex
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Co-authored-by: svlandeg <svlandeg@github.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Add experimental coref docs
* Docs cleanup
* Apply suggestions from code review
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Apply changes from code review
* Fix prettier formatting
It seems a period after a number made this think it was a list?
* Update docs on examples for initialize
* Add docs for coref scorers
* Remove 3.4 notes from coref
There won't be a "new" tag until it's in core.
* Add docs for span cleaner
* Fix docs
* Fix docs to match spacy-experimental
These weren't properly updated when the code was moved out of spacy
core.
* More doc fixes
* Formatting
* Update architectures
* Fix links
* Fix another link
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: svlandeg <svlandeg@github.com>
Preserve both `-` and `O` annotation in augmenters rather than relying
on `Example.to_dict`'s default support for one option outside of labeled
entity spans.
This is intended as a temporary workaround for augmenters for v3.4.x.
The behavior of `Example` and related IOB utils could be improved in the
general case for v3.5.
* Remove side effects from Doc.__init__()
* Changes based on review comment
* Readd test
* Change interface of Doc.__init__()
* Simplify test
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update doc.md
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update cupy extras:
* Extend to v11
* Add `cupy-cuda11x` and `cupy-wheel`
* Update quickstart to use `cupy-wheel` for CUDA 10.2+
* Rename cuda-wheel to cuda-autodetect, remove repeated CUDA in menu
* replicate bug with tok2vec in annotating components
* add overfitting test with a frozen tok2vec
* remove broadcast from predict and check doc.tensor instead
* remove broadcast
* proper error
* slight rephrase of documentation
* Enable Cython<->Python bindings for `Pipe` and `TrainablePipe` methods
* `pipes_with_nvtx_range`: Skip hooking methods whose signature cannot be ascertained
When loading pipelines from a config file, the arguments passed to individual pipeline components is validated by `pydantic` during init. For this, the validation model attempts to parse the function signature of the component's c'tor/entry point so that it can check if all mandatory parameters are present in the config file.
When using the `models_and_pipes_with_nvtx_range` as a `after_pipeline_creation` callback, the methods of all pipeline components get replaced by a NVTX range wrapper **before** the above-mentioned validation takes place. This can be problematic for components that are implemented as Cython extension types - if the extension type is not compiled with Python bindings for its methods, they will have no signatures at runtime. This resulted in `pydantic` matching the *wrapper's* parameters with the those in the config and raising errors.
To avoid this, we now skip applying the wrapper to any (Cython) methods that do not have signatures.
* new error message when 'project run assets'
* new error message when 'project run assets'
* Update spacy/cli/project/run.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Due to problems with the javascript conversion in the website
quickstart, remove the `has_letters` setting to simplify generating
`attrs` for the default `tok2vec`.
Additionally reduce `PREFIX` as in the trained pipelines.
* Add a dry run flag to download
* Remove --dry-run, add --url option to `spacy info` instead
* Make mypy happy
* Print only the URL, so it's easier to use in scripts
* Don't add the egg hash unless downloading an sdist
* Update spacy/cli/info.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Add two implementations of requirements
* Clean up requirements sample slightly
This should make mypy happy
* Update URL help string
* Remove requirements option
* Add url option to docs
* Add URL to spacy info model output, when available
* Add types-setuptools to testing reqs
* Add types-setuptools to requirements
* Add "compatible", expand docstring
* Update spacy/cli/info.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Run prettier on CLI docs
* Update docs
Add a sidebar about finding download URLs, with some examples of the new
command.
* Add download URLs to table on model page
* Apply suggestions from code review
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Updates from review
* download url -> download link
* Update docs
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* `Matcher`: Better type checking of values in `SetPredicate`
`SetPredicate`: Emit warning and return `False` on unexpected value types
* Rename `value_type_mismatch` variable
* Inline warning
* Remove unexpected type warning from `_SetPredicate`
* Ensure that `str` values are not interpreted as sequences
Check elements of sequence values for convertibility to `str` or `int`
* Add more `INTERSECT` and `IN` test cases
* Test for inputs with multiple characters
* Return `False` early instead of using a boolean flag
* Remove superfluous `int` check, parentheses
* Apply suggestions from code review
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Appy suggestions from code review
* Clarify test comment
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* adding unit test for spacy.load with disable/exclude string arg
* allow pure strings in from_config
* update docs
* upstream type adjustements
* docs update
* make docstring more consistent
* Update spacy/language.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* two more cleanups
* fix type in internal method
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Fix `test_{prefer,require}_gpu`
These tests assumed that GPUs are only supported with CuPy, but since Thinc 8.1
we also support Metal Performance Shaders.
* test_misc: arrange thinc imports to be together
* Add lang folder for la (Latin)
* Add Latin lang classes
* Add minimal tokenizer exceptions
* Add minimal stopwords
* Add minimal lex_attrs
* Update stopwords, tokenizer exceptions
* Add la tests; register la_tokenizer in conftest.py
* Update spacy/lang/la/lex_attrs.py
Remove duplicate form in Latin lex_attrs
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update natto-py version spec (#11222)
* Update natto-py version spec
* Update setup.cfg
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Add scorer to textcat API docs config settings (#11263)
* Update docs for pipeline initialize() methods (#11221)
* Update documentation for dependency parser
* Update documentation for trainable_lemmatizer
* Update documentation for entity_linker
* Update documentation for ner
* Update documentation for morphologizer
* Update documentation for senter
* Update documentation for spancat
* Update documentation for tagger
* Update documentation for textcat
* Update documentation for tok2vec
* Run prettier on edited files
* Apply similar changes in transformer docs
* Remove need to say annotated example explicitly
I removed the need to say "Must contain at least one annotated Example"
because it's often a given that Examples will contain some gold-standard
annotation.
* Run prettier on transformer docs
* chore: add 'concepCy' to spacy universe (#11255)
* chore: add 'concepCy' to spacy universe
* docs: add 'slogan' to concepCy
* Support full prerelease versions in the compat table (#11228)
* Support full prerelease versions in the compat table
* Fix types
* adding spans to doc_annotation in Example.to_dict (#11261)
* adding spans to doc_annotation in Example.to_dict
* to_dict compatible with from_dict: tuples instead of spans
* use strings for label and kb_id
* Simplify test
* Update data formats docs
Co-authored-by: Stefanie Wolf <stefanie.wolf@vitecsoftware.com>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Fix regex invalid escape sequences (#11276)
* Add W605 to the errors raised by flake8 in the CI (#11283)
* Clean up automated label-based issue handling (#11284)
* Clean up automated label-based issue handline
1. upgrade tiangolo/issue-manager to latest
2. move needs-more-info to tiangolo
3. change needs-more-info close time to 7 days
4. delete old needs-more-info config
* Use old, longer message
* Fix label name
* Fix Dutch noun chunks to skip overlapping spans (#11275)
* Add test for overlapping noun chunks
* Skip overlapping noun chunks
* Update spacy/tests/lang/nl/test_noun_chunks.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Docs: displaCy documentation - data types, `parse_{deps,ents,spans}`, spans example (#10950)
* add in spans example and parse references
* rm autoformatter
* rm extra ents copy
* TypedDict draft
* type fixes
* restore non-documentation files
* docs update
* fix spans example
* fix hyperlinks
* add parse example
* example fix + argument fix
* fix api arg in docs
* fix bad variable replacement
* fix spacing in style
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* fix spacing on table
* fix spacing on table
* rm temp files
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* include span_ruler for default warning filter (#11333)
* Add uk pipelines to website (#11332)
* Check for . in factory names (#11336)
* Make fixes for PR #11349
* Fix roman numeral coverage in #11349
Co-authored-by: Patrick J. Burns <patricks@diyclassics.org>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Paul O'Leary McCann <polm@dampfkraft.com>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Co-authored-by: Lj Miranda <12949683+ljvmiranda921@users.noreply.github.com>
Co-authored-by: Jules Belveze <32683010+JulesBelveze@users.noreply.github.com>
Co-authored-by: stefawolf <wlf.ste@gmail.com>
Co-authored-by: Stefanie Wolf <stefanie.wolf@vitecsoftware.com>
Co-authored-by: Peter Baumgartner <5107405+pmbaumgartner@users.noreply.github.com>
* Fix lookup usage (fix#11347)
Before using the lookups table in the French (and Catalan) lemmatizers,
there's a check to see if the current term is in the table. But it's
checking a string against hashes, so it's always false. Also the table
lookup function is designed so you don't have to do that anyway.
* Use the lookup table directly
* Use string, not token
* Add token and span custom attributes to to_json()
* Change logic for to_json
* Add functionality to from_json
* Small adjustments
* Move token/span attributes to new dict key
* Fix test
* Fix the same test but much better
* Add backwards compatibility tests and adjust logic
* Add test to check if attributes not set in underscore are not saved in the json
* Add tests for json compatibility
* Adjust test names
* Fix tests and clean up code
* Fix assert json tests
* small adjustment
* adjust naming and code readability
* Adjust naming, added more tests and changed logic
* Fix typo
* Adjust errors, naming, and small test optimization
* Fix byte tests
* Fix bytes tests
* Change naming and json structure
* update schema
* Update spacy/schemas.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update spacy/tokens/doc.pyx
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update spacy/tokens/doc.pyx
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update spacy/schemas.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update schema for underscore attributes
* Adjust underscore schema
* adjust schema tests
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Clean up automated label-based issue handline
1. upgrade tiangolo/issue-manager to latest
2. move needs-more-info to tiangolo
3. change needs-more-info close time to 7 days
4. delete old needs-more-info config
* Use old, longer message
* Fix label name
* adding spans to doc_annotation in Example.to_dict
* to_dict compatible with from_dict: tuples instead of spans
* use strings for label and kb_id
* Simplify test
* Update data formats docs
Co-authored-by: Stefanie Wolf <stefanie.wolf@vitecsoftware.com>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Added examples for Slovene
* Update spacy/lang/sl/examples.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Corrected a typo in one of the sentences
* Updated support for Slovenian
* Some minor changes to corrections
* Added forint currency
* Corrected HYPHENS_PERMITTED regex and some formatting
* Minor changes
* Un-xfail tokenizer test
* Format
Co-authored-by: Luka Dragar <D20124481@mytudublin.ie>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update documentation for dependency parser
* Update documentation for trainable_lemmatizer
* Update documentation for entity_linker
* Update documentation for ner
* Update documentation for morphologizer
* Update documentation for senter
* Update documentation for spancat
* Update documentation for tagger
* Update documentation for textcat
* Update documentation for tok2vec
* Run prettier on edited files
* Apply similar changes in transformer docs
* Remove need to say annotated example explicitly
I removed the need to say "Must contain at least one annotated Example"
because it's often a given that Examples will contain some gold-standard
annotation.
* Run prettier on transformer docs
* add additional REL_OP
* change to condition and new rel_op symbols
* add operators to docs
* add the anchor while we're in here
* add tests
Co-authored-by: Peter Baumgartner <5107405+pmbaumgartner@users.noreply.github.com>
After the precomputable affine table of shape [nB, nF, nO, nP] is
computed, padding with shape [1, nF, nO, nP] is assigned to the first
row of the precomputed affine table. However, when we are indexing the
precomputed table, we get a row of shape [nF, nO, nP]. CuPy versions
before 10.0 cannot paper over this shape difference.
This change fixes compatibility with CuPy < 10.0 by squeezing the first
dimension of the padding before assignment.
* precompute_hiddens/Parser: do not look up CPU ops
`get_ops("cpu")` is quite expensive. To avoid this, we want to cache the
result as in #11068. However, for 3.x we do not want to change the ABI.
So we avoid the expensive lookup by using NumpyOps. This should have a
minimal impact, since `get_ops("cpu")` was only used when the model ops
were `CupyOps`. If the ops are `AppleOps`, we are still passing through
the correct BLAS implementation.
* _NUMPY_OPS -> NUMPY_OPS
* `strings`: More roubust type checking of keys/IDs, coerce `int`-like types to `hash_t`
* Preserve existing public API behaviour
* Fix return type
* Replace `bool` with `bint`, rename to `_try_coerce_to_hash`, replace `id` with `hash`
* Avoid unnecessary re-encoding and re-calculation of strings and hashs respectively
* Rename variables named `hash`
Add comment on early return
* `TrainablePipe`: Add NVTX range decorator
* Annotate `TrainablePipe` subclasses with NVTX ranges
* Export function signature to allow introspection of args in tests
* Revert "Annotate `TrainablePipe` subclasses with NVTX ranges"
This reverts commit d8684f7372b2590bc603c3681a9679381253f8d6.
* Revert "Export function signature to allow introspection of args in tests"
This reverts commit f4405ca3ad710835e2861de0a846b8ec974718b0.
* Revert "`TrainablePipe`: Add NVTX range decorator"
This reverts commit 26536eb6b8508c71784a7606209c9a6664fb1b5e.
* Add `spacy.pipes_with_nvtx_range` pipeline callback
* Show warnings for all missing user-defined pipe functions that need to be annotated
Fix imports, typos
* Rename `DEFAULT_ANNOTATABLE_PIPE_METHODS` to `DEFAULT_NVTX_ANNOTATABLE_PIPE_METHODS`
Reorder import
* Walk model nodes directly whilst applying NVTX ranges
Ignore pipe method wrapper when applying range
* Add cuda116 and cuda117 extras
* Revert "remove `cuda116` extra from install widget (#11012)"
This reverts commit e7b498fb1f37261393180405aa3a3636ae57c709.
* Add cuda117 to quickstart
* Min_max_operators
1. Modified API and Usage for spaCy website to include min_max operator
2. Modified matcher.pyx to include min_max function {n,m} and its variants
3. Modified schemas.py to include min_max validation error
4. Added test cases to test_matcher_api.py, test_matcher_logic.py and test_pattern_validation.py
* attempt to fix mypy/pydantic compat issue
* formatting
* Update spacy/tests/matcher/test_pattern_validation.py
Co-authored-by: Source-Shen <82353723+Source-Shen@users.noreply.github.com>
Co-authored-by: svlandeg <svlandeg@github.com>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* vectors: avoid expensive comparisons between numpy ints and Python ints
* vectors: avoid failure on lists of ints
* Convert another numpy int to Python
Distinguish between vectors that are 0 vs. missing vectors when warning
about missing vectors.
Update `Doc.has_vector` to match `Span.has_vector` and
`Token.has_vector` for cases where the vocab has vectors but none of the
tokens in the container have vectors.
* Handle Russian, Ukrainian and Bulgarian
* Corrections
* Correction
* Correction to comment
* Changes based on review
* Correction
* Reverted irrelevant change in punctuation.py
* Remove unnecessary group
* Reverted accidental change
* Try cloning repo from main & master
* fixup! Try cloning repo from main & master
* fixup! fixup! Try cloning repo from main & master
* refactor clone and check for repo:branch existence
* spacing fix
* make mypy happy
* type util function
* Update spacy/cli/project/clone.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Peter Baumgartner <5107405+pmbaumgartner@users.noreply.github.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Use thinc-apple-ops>=0.1.0.dev0 with `apple` extras
Also test with thinc-apple-ops that is at least 0.1.0.dev0.
* Check thinc-apple-ops on macOS with Python 3.10
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Use `pip install --pre` for installing thinc-apple-ops in CI
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Enable flag on spacy.load: foundation for include, enable arguments.
* Enable flag on spacy.load: fixed tests.
* Enable flag on spacy.load: switched from pretrained model to empty model with added pipes for tests.
* Enable flag on spacy.load: switched to more consistent error on misspecification of component activity. Test refactoring. Added to default config.
* Enable flag on spacy.load: added support for fields not in pipeline.
* Enable flag on spacy.load: removed serialization fields from supported fields.
* Enable flag on spacy.load: removed 'enable' from config again.
* Enable flag on spacy.load: relaxed checks in _resolve_component_activation_status() to allow non-standard pipes.
* Enable flag on spacy.load: fixed relaxed checks for _resolve_component_activation_status() to allow non-standard pipes. Extended tests.
* Enable flag on spacy.load: comments w.r.t. resolution workarounds.
* Enable flag on spacy.load: remove include fields. Update website docs.
* Enable flag on spacy.load: updates w.r.t. changes in master.
* Implement Doc.from_json(): update docstrings.
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Implement Doc.from_json(): remove newline.
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Implement Doc.from_json(): change error message for E1038.
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Enable flag on spacy.load: wrapped docstring for _resolve_component_status() at 80 chars.
* Enable flag on spacy.load: changed exmples for enable flag.
* Remove newline.
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Fix docstring for Language._resolve_component_status().
* Rename E1038 to E1042.
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* account for NER labels with a hyphen in the name
* cleanup
* fix docstring
* add return type to helper method
* shorter method and few more occurrences
* user helper method across repo
* fix circular import
* partial revert to avoid circular import
The float -1 was returned rather than the integer -1 as the row for
unknown keys. This doesn't introduce a realy bug, since such floats
cast (without issues) to int in the conversion to NumPy arrays. Still,
it's nice to to do the correct thing :).
The `forward` of `precomputable_biaffine` performs matrix multiplication
and then `vstack`s the result with padding. This creates a temporary
array used for the output of matrix concatenation.
This change avoids the temporary by pre-allocating an array that is
large enough for the output of matrix multiplication plus padding and
fills the array in-place.
This gave me a small speedup (a bit over 100 WPS) on de_core_news_lg on
M1 Max (after changing thinc-apple-ops to support in-place gemm as BLIS
does).
* detect cycle during projectivize
* not complete test to detect cycle in projectivize
* boolean to int type to propagate error
* use unordered_set instead of set
* moved error message to errors
* removed cycle from test case
* use find instead of count
* cycle check: only perform one lookup
* Return bool again from _has_head_as_ancestor
Communicate presence of cycles through an output argument.
* Switch to returning std::pair to encode presence of a cycle
The has_cycle pointer is too easy to misuse. Ideally, we would have a
sum type like Rust's `Result` here, but C++ is not there yet.
* _is_non_proj_arc: clarify what we are returning
* _has_head_as_ancestor: remove count
We are now explicitly checking for cycles, so the algorithm must always
terminate. Either we encounter the head, we find a root, or a cycle.
* _is_nonproj_arc: simplify condition
* Another refactor using C++ exceptions
* Remove unused error code
* Print graph with cycle on exception
* Include .hh files in source package
* Add FIXME comment
* cycle detection test
* find cycle when starting from problematic vertex
Co-authored-by: Daniël de Kok <me@danieldk.eu>
* fix: De/Serialize `SpanGroups` including the SpanGroup keys
This prevents the loss of `SpanGroup`s that have the same .name as other `SpanGroup`s within the same `SpanGroups` object (upon de/serialization of the `SpanGroups`).
Fixes#10685
* Maintain backwards compatibility for serialized `SpanGroups`
(serialized as: a list of `SpanGroup`s, or b'')
* Add tests for `SpanGroups` deserialization backwards-compatibility
* Move a `SpanGroups` de/serialization test (test_issue10685)
to tests/serialize/test_serialize_spangroups.py
* Output a warning if deserializing a `SpanGroups` with duplicate .name-d `SpanGroup`s
* Minor refactor
* `SpanGroups.from_bytes` handles only `list` and `dict` types with
`dict` as the expected default
* For lists, keep first rather than last value encountered
* Update error message
* Rename and update tests
* Update to preserve list serialization of SpanGroups
To avoid breaking compatibility of serialized `Doc` and `DocBin` with
earlier versions of spacy v3, revert back to a list-only serialization,
but update the names just for serialization so that the SpanGroups keys
override the SpanGroup names.
* Preserve object identity and current key overwrite
* Preserve SpanGroup object identity
* Preserve last rather than first span group from SpanGroup list
format without SpanGroups keys
* Update inline comments
* Fix types
* Add type info for SpanGroup.copy
* Deserialize `SpanGroup`s as copies
when a single SpanGroup is the value for more than 1 `SpanGroups` key.
This is because we serialize `SpanGroups` as dicts (to maintain backward-
and forward-compatibility) and we can't assume `SpanGroup`s with the same
bytes/serialization were the same (identical) object, pre-serialization.
* Update spacy/tokens/_dict_proxies.py
* Add more SpanGroups serialization tests
Test that serialized SpanGroups maintain their Span order
* small clarification on older spaCy version
* Update spacy/tests/serialize/test_serialize_span_groups.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Add SpanRuler component
Add a `SpanRuler` component similar to `EntityRuler` that saves a list
of matched spans to `Doc.spans[spans_key]`. The matches from the token
and phrase matchers are deduplicated and sorted before assignment but
are not otherwise filtered.
* Update spacy/pipeline/span_ruler.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Fix cast
* Add self.key property
* Use number of patterns as length
* Remove patterns kwarg from init
* Update spacy/tests/pipeline/test_span_ruler.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Add options for spans filter and setting to ents
* Add `spans_filter` option as a registered function'
* Make `spans_key` optional and if `None`, set to `doc.ents` instead of
`doc.spans[spans_key]`.
* Update and generalize tests
* Add test for setting doc.ents, fix key property type
* Fix typing
* Allow independent doc.spans and doc.ents
* If `spans_key` is set, set `doc.spans` with `spans_filter`.
* If `annotate_ents` is set, set `doc.ents` with `ents_fitler`.
* Use `util.filter_spans` by default as `ents_filter`.
* Use a custom warning if the filter does not work for `doc.ents`.
* Enable use of SpanC.id in Span
* Support id in SpanRuler as Span.id
* Update types
* `id` can only be provided as string (already by `PatternType`
definition)
* Update all uses of Span.id/ent_id in Doc
* Rename Span id kwarg to span_id
* Update types and docs
* Add ents filter to mimic EntityRuler overwrite_ents
* Refactor `ents_filter` to take `entities, spans` args for more
filtering options
* Give registered filters more descriptive names
* Allow registered `filter_spans` filter
(`spacy.first_longest_spans_filter.v1`) to take any number of
`Iterable[Span]` objects as args so it can be used for spans filter
or ents filter
* Implement future entity ruler as span ruler
Implement a compatible `entity_ruler` as `future_entity_ruler` using
`SpanRuler` as the underlying component:
* Add `sort_key` and `sort_reverse` to allow the sorting behavior to be
customized. (Necessary for the same sorting/filtering as in
`EntityRuler`.)
* Implement `overwrite_overlapping_ents_filter` and
`preserve_existing_ents_filter` to support
`EntityRuler.overwrite_ents` settings.
* Add `remove_by_id` to support `EntityRuler.remove` functionality.
* Refactor `entity_ruler` tests to parametrize all tests to test both
`entity_ruler` and `future_entity_ruler`
* Implement `SpanRuler.token_patterns` and `SpanRuler.phrase_patterns`
properties.
Additional changes:
* Move all config settings to top-level attributes to avoid duplicating
settings in the config vs. `span_ruler/cfg`. (Also avoids a lot of
casting.)
* Format
* Fix filter make method name
* Refactor to use same error for removing by label or ID
* Also provide existing spans to spans filter
* Support ids property
* Remove token_patterns and phrase_patterns
* Update docstrings
* Add span ruler docs
* Fix types
* Apply suggestions from code review
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Move sorting into filters
* Check for all tokens in seen tokens in entity ruler filters
* Remove registered sort key
* Set Token.ent_id in a backwards-compatible way in Doc.set_ents
* Remove sort options from API docs
* Update docstrings
* Rename entity ruler filters
* Fix and parameterize scoring
* Add id to Span API docs
* Fix typo in API docs
* Include explicit labeled=True for scorer
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Fix TODO about typing
Fix was simple: just request an array2f.
* Add type ignore
Maxout has a more restrictive type than the residual layer expects (only
Floats2d vs any Floats).
* Various cleanup
This moves a lot of lines around but doesn't change any functionality.
Details:
1. use `continue` to reduce indentation
2. move sentence doc building inside conditional since it's otherwise
unused
3. reduces some temporary assignments
* Update docs for displacy style kwargs
Added "span" to the accepted values for the style kwarg in the displacy.serve and displacy.render top-level functions. These styles are new as of SpaCy 3.3, so I added the "new" tag for that option only
* restored alpha ordering
* Parser: use C saxpy/sgemm provided by the Ops implementation
This is a backport of https://github.com/explosion/spaCy/pull/10747
from the parser refactor branch. It eliminates the explicit calls
to BLIS, instead using the saxpy/sgemm provided by the Ops
implementation.
This allows us to use Accelerate in the parser on M1 Macs (with
an updated thinc-apple-ops).
Performance of the de_core_news_lg pipe:
BLIS 0.7.0, no thinc-apple-ops: 6385 WPS
BLIS 0.7.0, thinc-apple-ops: 36455 WPS
BLIS 0.9.0, no thinc-apple-ops: 19188 WPS
BLIS 0.9.0, thinc-apple-ops: 36682 WPS
This PR, thinc-apple-ops: 38726 WPS
Performance of the de_core_news_lg pipe (only tok2vec -> parser):
BLIS 0.7.0, no thinc-apple-ops: 13907 WPS
BLIS 0.7.0, thinc-apple-ops: 73172 WPS
BLIS 0.9.0, no thinc-apple-ops: 41576 WPS
BLIS 0.9.0, thinc-apple-ops: 72569 WPS
This PR, thinc-apple-ops: 87061 WPS
* Require thinc >=8.1.0,<8.2.0
* Lower thinc lowerbound to 8.1.0.dev0
* Use best CPU ops for CBLAS when the parser model is on the GPU
* Fix another unguarded cblas() call
* Fix: use ops as a shorthand for self.model.ops
Co-authored-by: Madeesh Kannan <shadeMe@users.noreply.github.com>
Co-authored-by: Madeesh Kannan <shadeMe@users.noreply.github.com>
This change removes `thinc.util.has_cupy` from the GPU presence check.
Currently `gpu_is_available` already implies `has_cupy`. We also want
to show this warning in the future when a machine has a non-CuPy GPU.
* Make changes to typing
* Correction
* Format with black
* Corrections based on review
* Bumped Thinc dependency version
* Bumped blis requirement
* Correction for older Python versions
* Update spacy/ml/models/textcat.py
Co-authored-by: Daniël de Kok <me@github.danieldk.eu>
* Corrections based on review feedback
* Readd deleted docstring line
Co-authored-by: Daniël de Kok <me@github.danieldk.eu>
* Add failing test
* Partial fix for issue
This kind of works. The issue with token length mismatches is gone. The
problem is that when you get empty lists of encodings to compare, it
fails because the sizes are not the same, even though they're both zero:
(0, 3) vs (0,). Not sure why that happens...
* Short circuit on empties
* Remove spurious check
The check here isn't needed now the the short circuit is fixed.
* Update spacy/tests/pipeline/test_entity_linker.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Use "eg", not "example"
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Rename to spans_key for consistency
* Implement spans length in debug data
* Implement how span bounds and spans are obtained
In this commit, I implemented how span boundaries (the tokens) around a
given span and spans are obtained. I've put them in the compile_gold()
function so that it's accessible later on. I will do the actual
computation of the span and boundary distinctiveness in the main
function above.
* Compute for p_spans and p_bounds
* Add computation for SD and BD
* Fix mypy issues
* Add weighted average computation
* Fix compile_gold conditional logic
* Add test for frequency distribution computation
* Add tests for kl-divergence computation
* Fix weighted average computation
* Make tables more compact by rounding them
* Add more descriptive checks for spans
* Modularize span computation methods
In this commit, I added the _get_span_characteristics and
_print_span_characteristics functions so that they can be reusable
anywhere.
* Remove unnecessary arguments and make fxs more compact
* Update a few parameter arguments
* Add tests for print_span and get_span methods
* Update API to talk about span characteristics in brief
* Add better reporting of spans_length
* Add test for span length reporting
* Update formatting of span length report
Removed '' to indicate that it's not a string, then
sort the n-grams by their length, not by their frequency.
* Apply suggestions from code review
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Show all frequency distribution when -V
In this commit, I displayed the full frequency distribution of the
span lengths when --verbose is passed. To make things simpler, I
rewrote some of the formatter functions so that I can call them
whenever.
Another notable change is that instead of showing percentages as
Integers, I showed them as floats (max 2-decimal places). I did this
because it looks weird when it displays (0%).
* Update logic on how total is computed
The way the 90% thresholding is computed now is that we keep
adding the percentages until we reach >= 90%. I also updated the wording
and used the term "At least" to denote that >= 90% of your spans have
these distributions.
* Fix display when showing the threshold percentage
* Apply suggestions from code review
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Add better phrasing for span information
* Update spacy/cli/debug_data.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Add minor edits for whitespaces etc.
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Add glossary entry for root
There was already one but it was lower case, maybe that should be
removed?
* remove lowercase root
On reflection, that was probably just a mistake.
* Add lowercase root back
It's harmless to leave it there.
* Pipe name override in config: added check with warning, added removal of name override from config, extended tests.
* Pipoe name override in config: added pytest UserWarning.
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Allow assets to be optional in spacy project: draft for optional flag/download_all options.
* Allow assets to be optional in spacy project: added OPTIONAL_DEFAULT reflecting default asset optionality.
* Allow assets to be optional in spacy project: renamed --all to --extra.
* Allow assets to be optional in spacy project: included optional flag in project config test.
* Allow assets to be optional in spacy project: added documentation.
* Allow assets to be optional in spacy project: fixing deprecated --all reference.
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Allow assets to be optional in spacy project: fixed project_assets() docstring.
* Allow assets to be optional in spacy project: adjusted wording in justification of optional assets.
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Allow assets to be optional in spacy project: switched to as keyword in project.yml. Updated docs.
* Allow assets to be optional in spacy project: updated comment.
* Allow assets to be optional in spacy project: replacing 'optional' with 'extra' in output.
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Allow assets to be optional in spacy project: replacing 'optional' with 'extra' in docstring..
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Allow assets to be optional in spacy project: replacing 'optional' with 'extra' in test..
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Allow assets to be optional in spacy project: replacing 'optional' with 'extra' in test.
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Allow assets to be optional in spacy project: renamed OPTIONAL_DEFAULT to EXTRA_DEFAULT.
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* add v1 and v2 tests for tok2vec architectures
* textcat architectures are not "layers"
* test older textcat architectures
* test older parser architecture
* Document different ways to create a pipeline: moved up/slightly modified paragraph on pipeline creation.
* Document different ways to create a pipeline: changed Finnish to Ukrainian in example for language without trained pipeline.
* Document different ways to create a pipeline: added explanation of blank pipeline.
* Document different ways to create a pipeline: exchanged Ukrainian with Yoruba.
* Fix StringStore.__getitem__ return type depending on parameter types
Small fix using `@overload` so that `StringStore.__getitem__` returns an `int` when given a `str` or `bytes` and a `str` when given an `int`.
* Update spacy/strings.pyi
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
The list of stop words for Spanish contained many inadequate words, see:
https://github.com/explosion/spaCy/issues/3052#issuecomment-1100760100
Removed words:
- verb forms of 'trabajar' (work) and intentar (try)
- words related to 'empleo' (employment)
- incorrect words: ampleamos, arribaabajo, soyos, paìs
- miscellaneous words due to being too significant of too infrequent:
actualmente, aproximadamente, antaño, cosas, ejemplo, horas, general,
pais, principalmente, raras
Added other stop words for completion:
- Spanish one-letter words
- numbers up to twelve
Some reformatting to 79 columns.
When in doubt, the English and German lists have been consulted as good
examples.
* `Matcher`: Remove superfluous GIL-acquiring check in `get_is_final`
This check incurred a significant performance penalty due to implict interactions between the GIL and Cython ref-counting code.
* `Matcher`: Inline `PatternStateC` accessors
* signing contributor agreement
* adding new content to the spaCy universe
* updating outdated example codes
* resolving issues for the PR
* resolve review for klayers
* remove contributor-agreement file from the PR
* Update code example of spaCySentiWS
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update spacy-sentiws code example
Co-authored-by: schaeran <schaeran1994@gmail.com>
Co-authored-by: schaeran <schaeran@explosion.ai>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Test for arc levels for identical arcs
Also moves the test in order with the other numbered tests.
* displaCy: filter identical arcs
Avoid increased levels due to identical arcs by first
filtering any identical arcs.
* Sort keys before filtering
Manual entry with keys out of order would previously become
different tuples and therefore not filtered correctly.
Co-authored-by: Joachim Fainberg <joachimfainberg@Joachims-MBP.lan>
* added crosslingual coreference to spacy universe
* Updated example to introduce batching example.
Co-authored-by: David Berenstein <david.berenstein@pandoraintelligence.com>
* Add basic tests for Tamil (ta)
* Add comment
Remove superfluous condition
* Remove superfluous call to `pipe`
Instantiate new tokenizer for special case
* Add initial design for diff command
For now, the diffing process looks like this:
- The default config is created based from some values in the user
config (e.g. which pipeline components were used, the lang, etc.)
- The user must supply manually if it was optimized for acc/efficiency
and if pretraining was involved.
* Make diff command structure similar to siblings
* Include gpu as a user option for CLI
* Make variables more explicit
* Fix type declaration for optimize enum
* Improve docstrings for diff CLI
* Add debug-diff to website API docs
* Switch position of configs so that user config is modded
* Add markdown flag for debug diff
This commit adds a --markdown (--md) flag that allows easier
copy-pasting to Github issues. Please note that this commit is dependent
on an unreleased version of wasabi (for the time being).
For posterity, the related PR is found here: https://github.com/ines/wasabi/pull/20
* Bump version of wasabi to 0.9.1
So that we can use the add_symbols parameter.
* Apply suggestions from code review
Co-authored-by: Ines Montani <ines@ines.io>
* Update docs based on code review suggestions
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Change command name from diff -> diff-config
* Clarify when options are relevant or not
* Rerun prettier on cli.md
Co-authored-by: Ines Montani <ines@ines.io>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Added test for overlapping arcs
* Provide distinct levels to overlapping arcs
* Update return type hint for get_levels
* Improved formatting spacy/displacy/render.py
Co-authored-by: Ines Montani <ines@ines.io>
Co-authored-by: Joachim Fainberg <joachimfainberg@Joachims-MacBook-Pro.local>
Co-authored-by: Ines Montani <ines@ines.io>
* Added new convenience cython functions to SpanGroup to avoid unnecessary allocation/deallocation of objects
* Replaced sorting in has_overlap with C++ for efficiency. Also, added a test for has_overlap
* Added a method to efficiently merge SpanGroups
* Added __delitem__, __add__ and __iadd__. Also, allowed to pass span lists to merge function. Replaced extend() body with call to merge
* Renamed merge to concat and added missing things to documentation
* Added operator+ and operator += in the documentation
* Added a test for Doc deallocation
* Update spacy/tokens/span_group.pyx
* Updated SpanGroup tests to use new span list comparison function rather than assert_span_list_equal, eliminating the need to have a separate assert_not_equal fnction
* Fixed typos in SpanGroup documentation
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Minor changes requested by Sofie: rearranged import statements. Added new=3.2.1 tag to SpanGroup.__setitem__ documentation
* SpanGroup: moved repetitive list index check/adjustment in a separate function
* Turn off formatting that hurts readability spacy/tests/doc/test_span_group.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Remove formatting that hurts readability spacy/tests/doc/test_span_group.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Turn off formatting that hurts readability in spacy/tests/doc/test_span_group.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Support more internal methods for SpanGroup
Add support for:
* `__setitem__`
* `__delitem__`
* `__iadd__`: for `SpanGroup` or `Iterable[Span]`
* `__add__`: for `SpanGroup` only
Adapted from #9698 with the scope limited to the magic methods.
* Use v3.3 as new version in docs
* Add new tag to SpanGroup.copy in API docs
* Remove duplicate import
* Apply suggestions from code review
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Remaining suggestions and formatting
Co-authored-by: nrodnova <nrodnova@hotmail.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Natalia Rodnova <4512370+nrodnova@users.noreply.github.com>
* Alignment: use a simplified ragged type for performance
This introduces the AlignmentArray type, which is a simplified version
of Ragged that performs better on the simple(r) indexing performed for
alignment.
* AlignmentArray: raise an error when using unsupported index
* AlignmentArray: move error messages to Errors
* AlignmentArray: remove simlified ... with simplifications
* AlignmentArray: fix typo that broke a[n:n] indexing
* added failing test case for the issue.
* Fixed typo.
* fixed typo in test.
* added corrected typo word into test_tr_lex_attrs_capitals as param. Test passes. Also tried and confirmed that test is failing after fixing the typo in the test case I wrote. Deleted the test case for typo.
Co-authored-by: Yunus Atahan <yunus.atahan@trmotor.local>
* Add vector deduplication
* Add `Vocab.deduplicate_vectors()`
* Always run deduplication in `spacy init vectors`
* Clean up a few vector-related error messages and docs examples
* Always unique with numpy
* Fix types
* Add edit tree lemmatizer
Co-authored-by: Daniël de Kok <me@danieldk.eu>
* Hide edit tree lemmatizer labels
* Use relative imports
* Switch to single quotes in error message
* Type annotation fixes
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Reformat edit_tree_lemmatizer with black
* EditTreeLemmatizer.predict: take Iterable
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Validate edit trees during deserialization
This change also changes the serialized representation. Rather than
mirroring the deep C structure, we use a simple flat union of the match
and substitution node types.
* Move edit_trees to _edit_tree_internals
* Fix invalid edit tree format error message
* edit_tree_lemmatizer: remove outdated TODO comment
* Rename factory name to trainable_lemmatizer
* Ignore type instead of casting truths to List[Union[Ints1d, Floats2d, List[int], List[str]]] for thinc v8.0.14
* Switch to Tagger.v2
* Add documentation for EditTreeLemmatizer
* docs: Fix 3.2 -> 3.3 somewhere
* trainable_lemmatizer documentation fixes
* docs: EditTreeLemmatizer is in edit_tree_lemmatizer.py
Co-authored-by: Daniël de Kok <me@danieldk.eu>
Co-authored-by: Daniël de Kok <me@github.danieldk.eu>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Added examples for Slovene
* Update spacy/lang/sl/examples.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Corrected a typo in one of the sentences
Co-authored-by: Luka Dragar <D20124481@mytudublin.ie>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update universe.json
added classy-classification to Spacy universe
* Update universe.json
added classy-classification to the spacy universe resources
* Update universe.json
corrected a small typo in json
* Update website/meta/universe.json
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update website/meta/universe.json
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update website/meta/universe.json
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update universe.json
processed merge feedback
* Update universe.json
* updated information for Classy Classificaiton
Made a more comprehensible and easy description for Classy Classification based on feedback of Philip Vollet to prepare for sharing.
* added note about examples
* corrected for wrong formatting changes
* Update website/meta/universe.json with small typo correction
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* resolved another typo
* Update website/meta/universe.json
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* added Concise Concepts package to spaCy universe.
* updated example code Concise Concepts
* updated description for Concise Concepts
* updated PR with more visually appealing examples
SO to koaning for the suggestions.
* corrected for small json typo's in concise concepts
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Add tokenizer option to allow Matcher handling for all rules
Add tokenizer option `with_faster_rules_heuristics` that determines
whether the special cases applied by the internal `Matcher` are filtered
by whether they contain affixes or space. If `True` (default), the rules
are filtered to prioritize speed over rare edge cases. If `False`, all
rules are included in the final `Matcher`-based pass over the doc.
* Reset all caches when reloading special cases
* Revert "Reset all caches when reloading special cases"
This reverts commit 4ef6bd171d00da01cbabc3bcde00088ba4bd5578.
* Initialize max_length properly
* Add new tag to API docs
* Rename to faster heuristics
* Save vectors as little endian, load with Ops.asarray
* Always save vector data as little endian
* Always run `Vectors.to_ops` when vector data is loaded so that
`Ops.asarray` can be used to load the data correctly for the current
ops.
* Update spacy/vectors.pyx
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update spacy/vectors.pyx
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Fix docstring for EntityRenderer
* Add warning in displacy if doc.spans are empty
* Implement parse_spans converter
One notable change here is that the default spans_key is sc, and
it's set by the user through the options.
* Implement SpanRenderer
Here, I implemented a SpanRenderer that looks similar to the
EntityRenderer except for some templates. The spans_key, by default, is
set to sc, but can be configured in the options (see parse_spans). The
way I rendered these spans is per-token, i.e., I first check if each
token (1) belongs to a given span type and (2) a starting token of a
given span type. Once I have this information, I render them into the
markup.
* Fix mypy issues on typing
* Add tests for displacy spans support
* Update colors from RGB to hex
Co-authored-by: Ines Montani <ines@ines.io>
* Remove unnecessary CSS properties
* Add documentation for website
* Remove unnecesasry scripts
* Update wording on the documentation
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Put typing dependency on top of file
* Put back z-index so that spans overlap properly
* Make warning more explicit for spans_key
Co-authored-by: Ines Montani <ines@ines.io>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update universe.json
added classy-classification to Spacy universe
* Update universe.json
added classy-classification to the spacy universe resources
* Update universe.json
corrected a small typo in json
* Update website/meta/universe.json
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update website/meta/universe.json
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update website/meta/universe.json
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update universe.json
processed merge feedback
* Update universe.json
* updated information for Classy Classificaiton
Made a more comprehensible and easy description for Classy Classification based on feedback of Philip Vollet to prepare for sharing.
* added note about examples
* corrected for wrong formatting changes
* Update website/meta/universe.json with small typo correction
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* resolved another typo
* Update website/meta/universe.json
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Tagger: use unnormalized probabilities for inference
Using unnormalized softmax avoids use of the relatively expensive exp function,
which can significantly speed up non-transformer models (e.g. I got a speedup
of 27% on a German tagging + parsing pipeline).
* Add spacy.Tagger.v2 with configurable normalization
Normalization of probabilities is disabled by default to improve
performance.
* Update documentation, models, and tests to spacy.Tagger.v2
* Move Tagger.v1 to spacy-legacy
* docs/architectures: run prettier
* Unnormalized softmax is now a Softmax_v2 option
* Require thinc 8.0.14 and spacy-legacy 3.0.9
* Add save_candidates attribute
* Change spancat api
* Add unit test
* reimplement method to produce a list of doc
* Add method to docs
* Add new version tag
* Add intended use to docstring
* prettier formatting
* Add support basic support for lower sorbian.
* Add some test for dsb.
* Update spacy/lang/dsb/examples.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Fix get_matching_ents
Not sure what happened here - the code prior to this commit simply does
not work. It's already covered by entity linker tests, which were
succeeding in the NEL PR, but couldn't possibly succeed on master.
* Fix test
Test was indented inside another test and so doesn't seem to have been
running properly.
* Add support basic support for upper sorbian.
* Add tokenizer exceptions and tests.
* Update spacy/lang/hsb/examples.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Partial fix of entity linker batching
* Add import
* Better name
* Add `use_gold_ents` option, docs
* Change to v2, create stub v1, update docs etc.
* Fix error type
Honestly no idea what the right type to use here is.
ConfigValidationError seems wrong. Maybe a NotImplementedError?
* Make mypy happy
* Add hacky fix for init issue
* Add legacy pipeline entity linker
* Fix references to class name
* Add __init__.py for legacy
* Attempted fix for loss issue
* Remove placeholder V1
* formatting
* slightly more interesting train data
* Handle batches with no usable examples
This adds a test for batches that have docs but not entities, and a
check in the component that detects such cases and skips the update step
as thought the batch were empty.
* Remove todo about data verification
Check for empty data was moved further up so this should be OK now - the
case in question shouldn't be possible.
* Fix gradient calculation
The model doesn't know which entities are not in the kb, so it generates
embeddings for the context of all of them.
However, the loss does know which entities aren't in the kb, and it
ignores them, as there's no sensible gradient.
This has the issue that the gradient will not be calculated for some of
the input embeddings, which causes a dimension mismatch in backprop.
That should have caused a clear error, but with numpyops it was causing
nans to happen, which is another problem that should be addressed
separately.
This commit changes the loss to give a zero gradient for entities not in
the kb.
* add failing test for v1 EL legacy architecture
* Add nasty but simple working check for legacy arch
* Clarify why init hack works the way it does
* Clarify use_gold_ents use case
* Fix use gold ents related handling
* Add tests for no gold ents and fix other tests
* Use aligned ents function (not working)
This doesn't actually work because the "aligned" ents are gold-only. But
if I have a different function that returns the intersection, *then*
this will work as desired.
* Use proper matching ent check
This changes the process when gold ents are not used so that the
intersection of ents in the pred and gold is used.
* Move get_matching_ents to Example
* Use model attribute to check for legacy arch
* Rename flag
* bump spacy-legacy to lower 3.0.9
Co-authored-by: svlandeg <svlandeg@github.com>
* fixing argument order for rehearse
* rehearse test for ner and tagger
* rehearse bugfix
* added test for parser
* test for multilabel textcat
* rehearse fix
* remove debug line
* Update spacy/tests/training/test_rehearse.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update spacy/tests/training/test_rehearse.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Kádár Ákos <akos@onyx.uvt.nl>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Make core projectivization methods cdef nogil
While profiling the parser, I noticed that relatively a lot of time is
spent in projectivization. This change rewrites the functions in the
core loops as cdef nogil for efficiency.
In C++-land, we use vector in place of Python lists and absent heads
are represented as -1 in place of None.
* _heads_to_c: add assertion
Validation should be performed by the caller, but this assertion ensures that
we are not reading/writing out of bounds with incorrect input.
* Fix NER check in CoNLL-U converter
Leave ents unset if no NER annotation is found in the MISC column.
* Revert to global rather than per-sentence NER check
* Update spacy/training/converters/conllu_to_docs.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Add whitespace augmenter that inserts a single whitespace token into a
doc containing annotation used in core trained pipelines.
Add a combined augmenter that handles lowercasing, orth variants and
whitespace augmentation.
* Extended list of numbers for ru language
Extended list of numbers with all forms and cases including short forms, slang variants and roman numerals.
* Update lex_attrs.py
* Update 'like_num' function with percentages
Added support for numbers with percentages like 12%, 1.2% and etc. to the 'like_num' function.
* black formatting
Co-authored-by: thomashacker <EdwardSchmuhl@web.de>
* Extend list of abbreviations for ru language
Extended list of abbreviations for ru language those may have influence on tokenization.
* black formatting
Co-authored-by: thomashacker <EdwardSchmuhl@web.de>
* Delay loading of mecab in Korean tokenizer
Delay loading of mecab until the tokenizer is called the first time so
that it's possible to initialize a blank `ko` pipeline without having
mecab installed, e.g. for use with `spacy init vectors`.
* Move mecab import back to __init__
Move mecab import back to __init__ to warn users at the same point as
before for missing python dependencies.
* Add github actions for slow and gpu tests
* change weekly GPU tests to also run slow tests, and change the time
* only run the tests if there were commits in the past day
* remove duplicate line
* add sent start/end token attributes to the docs
* let has_annotation work with IS_SENT_END
* elif instead of if
* add has_annotation test for sent attributes
* fix typo
* remove duplicate is_sent_start entry in docs
* Setup debug data for spancat
* Add check for missing labels
* Add low-level data warning error
* Improve logic when compiling the gold train data
* Implement check for negative examples
* Remove breakpoint
* Remove ws_ents and missing entity checks
* Fix mypy errors
* Make variable name spans_key consistent
* Rename pipeline -> component for consistency
* Account for missing labels per spans_key
* Cleanup variable names for consistency
* Improve brevity of conditional statements
* Remove unused variables
* Include spans_key as an argument for _get_examples
* Add a conditional check for spans_key
* Update spancat debug data based on new API
- Instead of using _get_labels_from_model(), I'm now using
_get_labels_from_spancat() (cf. https://github.com/explosion/spaCy/pull10079)
- The way information is displayed was also changed (text -> table)
* Rename model_labels to ensure mypy works
* Update wording on warning messages
Use "span type" instead of "entity type" in wording the warning messages.
This is because Spans aren't necessarily entities.
* Update component type into a Literal
This is to make it clear that the component parameter should only accept
either 'spancat' or 'ner'.
* Update checks to include actual model span_keys
Instead of looking at everything in the data, we only check those
span_keys from the actual spancat component. Instead of doing the filter
inside the for-loop, I just made another dictionary,
data_labels_in_component to hold this value.
* Update spacy/cli/debug_data.py
* Show label counts only when verbose is True
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Fix debug data check for ents that cross sents
* Use aligned sent starts to have the same indices for the NER and sent
start annotation
* Add a temporary, insufficient hack for the case where a
sentence-initial reference token is split into multiple tokens in the
predicted doc, since `Example.get_aligned("SENT_START")` currently
aligns `True` to all the split tokens.
* Improve test example
* Use Example.get_aligned_sent_starts
* Add test for crossing entity
* Auto-format code with black
* add black requirement to dev dependencies and pin to 22.x
* ignore black dependency for comparison with setup.cfg
Co-authored-by: explosion-bot <explosion-bot@users.noreply.github.com>
Co-authored-by: svlandeg <svlandeg@github.com>
So that overriding `paths.vectors` works consistently in generated
configs, set vectors model in `paths.vectors` and always refer to this
path in `initialize.vectors`.
Remove exception for whitespace tokens in `Example.get_aligned` so that
annotation on whitespace tokens is aligned in the same way as for
non-whitespace tokens.
* Added spacy-wrap to universe
Added spacy-wrap to universe a small package for wrapping fine-tuned huggingface transformers to a spacy pipeline following the same API as spacy-transformers. (Currently limited to classification models)
* Update website/meta/universe.json
* Update website/meta/universe.json
* Update website/meta/universe.json
* Update website/meta/universe.json
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Clarify Span.ents documentation
Ref: #10135
Retain current behaviour. Span.ents will only include entities within
said span. You can't get tokens outside of the original span.
* Reword docstrings
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update API docs in the website
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* This comma has been most probably been left out unintentionally, leading to string concatenation between the two consecutive lines. This issue has been found automatically using a regular expression.
* This comma has been most probably been left out unintentionally, leading to string concatenation between the two consecutive lines. This issue has been found automatically using a regular expression.
* Fix infix as prefix in Tokenizer.explain
Update `Tokenizer.explain` to align with the `Tokenizer` algorithm:
* skip infix matches that are prefixes in the current substring
* Update tokenizer pseudocode in docs
* Improve typing hints for Matcher.__call__
* Add typing hints for DependencyMatcher
* Add typing hints to underscore extensions
* Update Doc.tensor type (requires numpy 1.21)
* Fix typing hints for Language.component decorator
* Use generic np.ndarray type in Doc to avoid numpy version update
* Fix mypy errors
* Fix cyclic import caused by Underscore typing hints
* Use Literal type from spacy.compat
* Update matcher.pyi import format
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Instead of the running the actual suggester, which may require
annotation from annotating components that is not necessarily present in
the reference docs, use the built-in 1-gram suggester.
* added iob to int
* added tests
* added iob strings
* added error
* blacked attrs
* Update spacy/tests/lang/test_attrs.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update spacy/attrs.pyx
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* added iob strings as global
* minor refinement with iob
* removed iob strings from token
* changed to uppercase
* cleaned and went back to master version
* imported iob from attrs
* Update and format errors
* Support and test both str and int ENT_IOB key
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* added new field
* added exception for IOb strings
* minor refinement to schema
* removed field
* fixed typo
* imported numeriacla val
* changed the code bit
* cosmetics
* added test for matcher
* set ents of moc docs
* added invalid pattern
* minor update to documentation
* blacked matcher
* added pattern validation
* add IOB vals to schema
* changed into test
* mypy compat
* cleaned left over
* added compat import
* changed type
* added compat import
* changed literal a bit
* went back to old
* made explicit type
* Update spacy/schemas.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update spacy/schemas.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update spacy/schemas.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Determine labels by factory name in debug data
For all components, return labels for all components with the
corresponding factory name rather than for only the default name.
For `spancat`, return labels as a dict keyed by `spans_key`.
* Refactor for typing
* Add test
* Use assert instead of cast, removed unneeded arg
* Mark test as slow
* Add link to pattern file info in EntityRuler.initialize docs
* Update website/docs/api/entityruler.md
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Use Vectors.shape rather than Vectors.data.shape
* Use Vectors.size rather than Vectors.data.size
* Add Vectors.to_ops to move data between different ops
* Add documentation for Vector.to_ops
By @polm, redone from #9917 after incorrect (reverted) rebase.
`sudachipy>=0.5.2` is needed for newer dictionaries. `sudachipy<0.6.0`
is kept for users who might still prefer the older version, in
particular to be able to compile it without rust.
* Corrected Span's __richcmp__ implementation to take end, label and kb_id in consideration
* Updated test
* Updated test
* Removed formatting from a test for readability sake
* Use same tuples for all comparisons
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* add entry for Applied Language Technology under "Courses"
Added the following entry into `universe.json`:
```
{
"type": "education",
"id": "applt-course",
"title": "Applied Language Technology",
"slogan": "NLP for newcomers using spaCy and Stanza",
"description": "These learning materials provide an introduction to applied language technology for audiences who are unfamiliar with language technology and programming. The learning materials assume no previous knowledge of the Python programming language.",
"url": "https://applied-language-technology.readthedocs.io/",
"image": "https://www.mv.helsinki.fi/home/thiippal/images/applt-preview.jpg",
"thumb": "https://applied-language-technology.readthedocs.io/en/latest/_static/logo.png",
"author": "Tuomo Hiippala",
"author_links": {
"twitter": "tuomo_h",
"github": "thiippal",
"website": "https://www.mv.helsinki.fi/home/thiippal/"
},
"category": ["courses"]
},
```
* Update the entry for "Applied Language Technology"
* Edited Slovenian stop words list (#9707)
* Noun chunks for Italian (#9662)
* added it vocab
* copied portuguese
* added possessive determiner
* added conjed Nps
* added nmoded Nps
* test misc
* more examples
* fixed typo
* fixed parenth
* fixed comma
* comma fix
* added syntax iters
* fix some index problems
* fixed index
* corrected heads for test case
* fixed tets case
* fixed determiner gender
* cleaned left over
* added example with apostophe
* French NP review (#9667)
* adapted from pt
* added basic tests
* added fr vocab
* fixed noun chunks
* more examples
* typo fix
* changed naming
* changed the naming
* typo fix
* Add Japanese kana characters to default exceptions (fix#9693) (#9742)
This includes the main kana, or phonetic characters, used in Japanese.
There are some supplemental kana blocks in Unicode outside the BMP that
could also be included, but because their actual use is rare I omitted
them for now, but maybe they should be added. The omitted blocks are:
- Kana Supplement
- Kana Extended (A and B)
- Small Kana Extension
* Remove NER words from stop words in Norwegian (#9820)
Default stop words in Norwegian bokmål (nb) in Spacy contain important entities, e.g. France, Germany, Russia, Sweden and USA, police district, important units of time, e.g. months and days of the week, and organisations.
Nobody expects their presence among the default stop words. There is a danger of users complying with the general recommendation of filtering out stop words, while being unaware of filtering out important entities from their data.
See explanation in https://github.com/explosion/spaCy/issues/3052#issuecomment-986756711 and comment https://github.com/explosion/spaCy/issues/3052#issuecomment-986951831
* Bump sudachipy version
* Update sudachipy versions
* Bump versions
Bumping to the most recent dictionary just to keep thing current.
Bumping sudachipy to 5.2 because older versions don't support recent
dictionaries.
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Richard Hudson <richard@explosion.ai>
Co-authored-by: Duygu Altinok <duygu@explosion.ai>
Co-authored-by: Haakon Meland Eriksen <haakon.eriksen@far.no>
This change changes the type of left/right-arc collections from
vector[ArcC] to unordered_map[int, vector[Arc]], so that the arcs are
keyed by the head. This allows us to find all the left/right arcs for a
particular head in constant time in StateC::{L,R}.
Benchmarks with long docs (N is the number of text repetitions):
Before (using #10019):
N Time (s)
400 3.2
800 5.0
1600 9.5
3200 23.2
6400 66.8
12800 220.0
After (this commit):
N Time (s)
400 3.1
800 4.3
1600 6.7
3200 12.0
6400 22.0
12800 42.0
Related to #9858 and #10019.
* Speed up the StateC::L feature function
This function gets the n-th most-recent left-arc with a particular head.
Before this change, StateC::L would construct a vector of all left-arcs
with the given head and then pick the n-th most recent from that vector.
Since the number of left-arcs strongly correlates with the doc length
and the feature is constructed for every transition, this can make
transition-parsing quadratic.
With this change StateC::L:
- Searches left-arcs backwards.
- Stops early when the n-th matching transition is found.
- Does not construct a vector (reducing memory pressure).
This change doesn't avoid the linear search when the transition that is
queried does not occur in the left-arcs. Regardless, performance is
improved quite a bit with very long docs:
Before:
N Time
400 3.3
800 5.4
1600 11.6
3200 30.7
After:
N Time
400 3.2
800 5.0
1600 9.5
3200 23.2
We can probably do better with more tailored data structures, but I
first wanted to make a low-impact PR.
Found while investigating #9858.
* StateC::L: simplify loop
* Speed up the StateC::L feature function
This function gets the n-th most-recent left-arc with a particular head.
Before this change, StateC::L would construct a vector of all left-arcs
with the given head and then pick the n-th most recent from that vector.
Since the number of left-arcs strongly correlates with the doc length
and the feature is constructed for every transition, this can make
transition-parsing quadratic.
With this change StateC::L:
- Searches left-arcs backwards.
- Stops early when the n-th matching transition is found.
- Does not construct a vector (reducing memory pressure).
This change doesn't avoid the linear search when the transition that is
queried does not occur in the left-arcs. Regardless, performance is
improved quite a bit with very long docs:
Before:
N Time
400 3.3
800 5.4
1600 11.6
3200 30.7
After:
N Time
400 3.2
800 5.0
1600 9.5
3200 23.2
We can probably do better with more tailored data structures, but I
first wanted to make a low-impact PR.
Found while investigating #9858.
* StateC::L: simplify loop
* Fix Scorer.score_cats for missing labels
* Add test case for Scorer.score_cats missing labels
* semantic nitpick
* black formatting
* adjust test to give different results depending on multi_label setting
* fix loss function according to whether or not missing values are supported
* add note to docs
* small fixes
* make mypy happy
* Update spacy/pipeline/textcat.py
Co-authored-by: Florian Cäsar <florian.caesar@pm.me>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: svlandeg <svlandeg@github.com>
* change '_' to '' to allow Token.pos, when no value for token pos in conllu data
* Minor code style
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Default stop words in Norwegian bokmål (nb) in Spacy contain important entities, e.g. France, Germany, Russia, Sweden and USA, police district, important units of time, e.g. months and days of the week, and organisations.
Nobody expects their presence among the default stop words. There is a danger of users complying with the general recommendation of filtering out stop words, while being unaware of filtering out important entities from their data.
See explanation in https://github.com/explosion/spaCy/issues/3052#issuecomment-986756711 and comment https://github.com/explosion/spaCy/issues/3052#issuecomment-986951831
* added ruler coe
* added error for none existing pattern
* changed error to warning
* changed error to warning
* added basic tests
* fixed place
* added test files
* went back to error
* went back to pattern error
* minor change to docs
* changed style
* changed doc
* changed error slightly
* added remove to phrasem api
* error key already existed
* phrase matcher match code to api
* blacked tests
* moved comments before expr
* corrected error no
* Update website/docs/api/entityruler.md
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update website/docs/api/entityruler.md
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Added sents property to Span class that returns a generator of sentences the Span belongs to
* Added description to Span.sents property
* Update test_span to clarify the difference between span.sent and span.sents
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update spacy/tests/doc/test_span.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Fix documentation typos in spacy/tokens/span.pyx
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update Span.sents doc string in spacy/tokens/span.pyx
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Parametrized test_span_spans
* Corrected Span.sents to check for span-level hook first. Also, made Span.sent respect doc-level sents hook if no span-level hook is provided
* Corrected Span ocumentation copy/paste issue
* Put back accidentally deleted lines
* Fixed formatting in span.pyx
* Moved check for SENT_START annotation after user hooks in Span.sents
* add version where the property was introduced
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Migrate regressions 1-1000
* Move serialize test to correct file
* Remove tests that won't work in v3
* Migrate regressions 1000-1500
Removed regression test 1250 because v3 doesn't support the old LEX
scheme anymore.
* Add missing imports in serializer tests
* Migrate tests 1500-2000
* Migrate regressions from 2000-2500
* Migrate regressions from 2501-3000
* Migrate regressions from 3000-3501
* Migrate regressions from 3501-4000
* Migrate regressions from 4001-4500
* Migrate regressions from 4501-5000
* Migrate regressions from 5001-5501
* Migrate regressions from 5501 to 7000
* Migrate regressions from 7001 to 8000
* Migrate remaining regression tests
* Fixing missing imports
* Update docs with new system [ci skip]
* Update CONTRIBUTING.md
- Fix formatting
- Update wording
* Remove lemmatizer tests in el lang
* Move a few tests into the general tokenizer
* Separate Doc and DocBin tests
This includes the main kana, or phonetic characters, used in Japanese.
There are some supplemental kana blocks in Unicode outside the BMP that
could also be included, but because their actual use is rare I omitted
them for now, but maybe they should be added. The omitted blocks are:
- Kana Supplement
- Kana Extended (A and B)
- Small Kana Extension
* morphologizer: avoid recreating label tuple for each token
The `labels` property converts the dictionary key set to a tuple. This
property was used for every annotated token, recreating the tuple over
and over again.
Construct the tuple once in the set_annotations function and reuse it.
On a Finnish pipeline that I was experimenting with, this results in a
speedup of ~15% (~13000 -> ~15000 WPS).
* tagger: avoid recreating label tuple for each token
* Add support for kb_id to be displayed via displacy.serve. The current support is only limited to the manual option in displacy.render
* Commit to check pre-commit hooks are run.
* Update spacy/displacy/__init__.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Changes as per suggestions on the PR.
* Update website/docs/api/top-level.md
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update website/docs/api/top-level.md
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* tag option as new from 3.2.1 onwards
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com>
* Use internal names for factories
If a component factory is registered like `@French.factory(...)` instead
of `@Language.factory(...)`, the name in the factories registry will be
prefixed with the language code. However in the nlp.config object the
factory will be listed without the language code. The `add_pipe` code
has fallback logic to handle this, but packaging code and the registry
itself don't.
This change makes it so that the factory name in nlp.config is the
language-specific form. It's not clear if this will break anything else,
but it does seem to fix the inconsistency and resolve the specific user
issue that brought this to our attention.
* Change approach to use fallback in package lookup
This adds fallback logic to the package lookup, so it doesn't have to
touch the way the config is built. It seems to fix the tests too.
* Remove unecessary line
* Add test
Thsi also adds an assert that seems to have been forgotten.
* Added Slovak
* Added Slovenian tests
* Added Estonian tests
* Added Croatian tests
* Added Latvian tests
* Added Icelandic tests
* Added Afrikaans tests
* Added language-independent tests
* Added Kannada tests
* Tidied up
* Added Albanian tests
* Formatted with black
* Added failing tests for anomalies
* Update spacy/tests/lang/af/test_text.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Added context to failing Estonian tokenizer test
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Added context to failing Croatian tokenizer test
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Added context to failing Icelandic tokenizer test
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Added context to failing Latvian tokenizer test
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Added context to failing Slovak tokenizer test
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Added context to failing Slovenian tokenizer test
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Added ENT_ID and ENT_KB_ID into the list of the attributes that Matcher matches on
* Added ENT_ID and ENT_KB_ID to TEST_PATTERNS in test_pattern_validation.py. Disabled tests that I added before
* Update website/docs/api/matcher.md
* Format
* Remove skipped tests
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* added error string
* added serialization test
* added more to if statements
* wrote file to tempdir
* added tempdir
* changed parameter a bit
* Update spacy/tests/pipeline/test_entity_ruler.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
If the predicted docs are missing annotation according to
`has_annotation`, treat the docs as having no predictions rather than
raising errors when the annotation is missing.
The motivation for this is a combined tokenization+sents scorer for a
component where the sents annotation is optional. To provide a single
scorer in the component factory, it needs to be possible for the scorer
to continue despite missing sents annotation in the case where the
component is not annotating sents.
* Add note on batch contract
Using listeners requires batches to be consistent. This is obvious if
you understand how the listener works, but it wasn't clearly stated in
the Docs, and was subtle enough that the EntityLinker missed it.
There is probably a clearer way to explain what the actual requirement
is, but I figure this is a good start.
* Rewrite to clarify role of caching
Exclude strings from `Vector.to_bytes()` comparions for v3.2+ `Vectors`
that now include the string store so that the source vector comparison
is only comparing the vectors and not the strings.
* Clarify how to fill in init_tok2vec after pretraining
* Ignore init_tok2vec arg in pretraining
* Update docs, config setting
* Remove obsolete note about not filling init_tok2vec early
This seems to have also caught some lines that needed cleanup.
* make nlp.pipe() return None docs when no exceptions are (re-)raised during error handling
* Remove changes other than as_tuples test
* Only check warning count for one process
* Fix types
* Format
Co-authored-by: Xi Bai <xi.bai.ed@gmail.com>
* Add section for spacy.cli.train.train
* Add link from training page to train function
* Ensure path in train helper
* Update docs
Co-authored-by: Ines Montani <ines@ines.io>
* Add micro PRF for morph scoring
For pipelines where morph features are added by more than one component
and a reference training corpus may not contain all features, a micro
PRF score is more flexible than a simple accuracy score. An example is
the reading and inflection features added by the Japanese tokenizer.
* Use `morph_micro_f` as the default morph score for Japanese
morphologizers.
* Update docstring
* Fix typo in docstring
* Update Scorer API docs
* Fix results type
* Organize score list by attribute prefix
* Update for python 3.10
* Update mac image
* Update build constraints for python 3.10
* Add extras for cupy cuda 11.3-11.5
* Remove cupy-cuda115 extra
* Require thinc>=8.0.12
* Switch CI to windows-2019
* Skip mypy for python 3.10
So that the install/upgrade quickstart also upgrades
`spacy-transformers` with `pip install spacy[transformers]`, require
`spacy-transformers>=1.1.2` in the `transformers` extra.
* Add support for fasttext-bloom hash-only vectors
Overview:
* Extend `Vectors` to have two modes: `default` and `ngram`
* `default` is the default mode and equivalent to the current
`Vectors`
* `ngram` supports the hash-only ngram tables from `fasttext-bloom`
* Extend `spacy.StaticVectors.v2` to handle both modes with no changes
for `default` vectors
* Extend `spacy init vectors` to support ngram tables
The `ngram` mode **only** supports vector tables produced by this
fork of fastText, which adds an option to represent all vectors using
only the ngram buckets table and which uses the exact same ngram
generation algorithm and hash function (`MurmurHash3_x64_128`).
`fasttext-bloom` produces an additional `.hashvec` table, which can be
loaded by `spacy init vectors --fasttext-bloom-vectors`.
https://github.com/adrianeboyd/fastText/tree/feature/bloom
Implementation details:
* `Vectors` now includes the `StringStore` as `Vectors.strings` so that
the API can stay consistent for both `default` (which can look up from
`str` or `int`) and `ngram` (which requires `str` to calculate the
ngrams).
* In ngram mode `Vectors` uses a default `Vectors` object as a cache
since the ngram vectors lookups are relatively expensive.
* The default cache size is the same size as the provided ngram vector
table.
* Once the cache is full, no more entries are added. The user is
responsible for managing the cache in cases where the initial
documents are not representative of the texts.
* The cache can be resized by setting `Vectors.ngram_cache_size` or
cleared with `vectors._ngram_cache.clear()`.
* The API ends up a bit split between methods for `default` and for
`ngram`, so functions that only make sense for `default` or `ngram`
include warnings with custom messages suggesting alternatives where
possible.
* `Vocab.vectors` becomes a property so that the string stores can be
synced when assigning vectors to a vocab.
* `Vectors` serializes its own config settings as `vectors.cfg`.
* The `Vectors` serialization methods have added support for `exclude`
so that the `Vocab` can exclude the `Vectors` strings while serializing.
Removed:
* The `minn` and `maxn` options and related code from
`Vocab.get_vector`, which does not work in a meaningful way for default
vector tables.
* The unused `GlobalRegistry` in `Vectors`.
* Refactor to use reduce_mean
Refactor to use reduce_mean and remove the ngram vectors cache.
* Rename to floret
* Rename to floret in error messages
* Use --vectors-mode in CLI, vector init
* Fix vectors mode in init
* Remove unused var
* Minor API and docstrings adjustments
* Rename `--vectors-mode` to `--mode` in `init vectors` CLI
* Rename `Vectors.get_floret_vectors` to `Vectors.get_batch` and support
both modes.
* Minor updates to Vectors docstrings.
* Update API docs for Vectors and init vectors CLI
* Update types for StaticVectors
* Ignore prefix in suffix matches
Ignore the currently matched prefix when looking for suffix matches in
the tokenizer. Otherwise a lookbehind in the suffix pattern may match
incorrectly due the presence of the prefix in the token string.
* Move °[cfkCFK]. to a tokenizer exception
* Adjust exceptions for same tokenization as v3.1
* Also update test accordingly
* Continue to split . after °CFK if ° is not a prefix
* Exclude new ° exceptions for pl
* Switch back to default tokenization of "° C ."
* Revert "Exclude new ° exceptions for pl"
This reverts commit 952013a5b4114ca0ed3b65285f50e8ef05c1695a.
* Add exceptions for °C for hu
* Raise an error when multiprocessing is used on a GPU
As reported in #5507, a confusing exception is thrown when
multiprocessing is used with a GPU model and the `fork` multiprocessing
start method:
cupy.cuda.runtime.CUDARuntimeError: cudaErrorInitializationError: initialization error
This change checks whether one of the models uses the GPU when
multiprocessing is used. If so, raise a friendly error message.
Even though multiprocessing can work on a GPU with the `spawn` method,
it quickly runs the GPU out-of-memory on real-world data. Also,
multiprocessing on a single GPU typically does not provide large
performance gains.
* Move GPU multiprocessing check to Language.pipe
* Warn rather than error when using multiprocessing with GPU models
* Improve GPU multiprocessing warning message.
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Reduce API assumptions
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update spacy/language.py
* Update spacy/language.py
* Test that warning is thrown with GPU + multiprocessing
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* add custom protocols in spacy.ty
* add a test for the new types in spacy.ty
* import Example when type checking
* some type fixes
* put Protocol in compat
* revert update check back to hasattr
* runtime_checkable in compat as well
* Replace use_ops("numpy") by use_ops("cpu") in the parser
This ensures that the best available CPU implementation is chosen
(e.g. Thinc Apple Ops on macOS).
* Run spaCy tests with apple-thinc-ops on macOS
* Remove some old version refs in the docs
* Remove warning
* Update spacy/matcher/matcher.pyx
* Remove all references to the punctuation warning
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Add the spacy.models_with_nvtx_range.v1 callback
This callback recursively adds NVTX ranges to the Models in each pipe in
a pipeline.
* Fix create_models_with_nvtx_range type signature
* NVTX range: wrap models of all trainable pipes jointly
This avoids that (sub-)models that are shared between pipes get wrapped
twice.
* NVTX range callback: make color configurable
Add forward_color and backprop_color options to set the color for the
NVTX range.
* Move create_models_with_nvtx_range to spacy.ml
* Update create_models_with_nvtx_range for thinc changes
with_nvtx_range now updates an existing node, rather than returning a
wrapper node. So, we can simply walk over the nodes and update them.
* NVTX: use after_pipeline_creation in example
* Add note about how the model name is used
* Add link to TransformersModel docs, separate paragraph
* Local link
* Revise docs
* Update website/docs/usage/embeddings-transformers.md
* Update website/docs/usage/embeddings-transformers.md
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Add note about how the model name is used
* Add link to TransformersModel docs, separate paragraph
* Local link
* Revise docs
* Update website/docs/usage/embeddings-transformers.md
* Update website/docs/usage/embeddings-transformers.md
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* add examples and num_words
* add contributor agreement
* Update spacy/lang/vi/examples.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* consistent format
add empty line at the end of file
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* 🚨 Ignore all existing Mypy errors
* 🏗 Add Mypy check to CI
* Add types-mock and types-requests as dev requirements
* Add additional type ignore directives
* Add types packages to dev-only list in reqs test
* Add types-dataclasses for python 3.6
* Add ignore to pretrain
* 🏷 Improve type annotation on `run_command` helper
The `run_command` helper previously declared that it returned an
`Optional[subprocess.CompletedProcess]`, but it isn't actually possible
for the function to return `None`. These changes modify the type
annotation of the `run_command` helper and remove all now-unnecessary
`# type: ignore` directives.
* 🔧 Allow variable type redefinition in limited contexts
These changes modify how Mypy is configured to allow variables to have
their type automatically redefined under certain conditions. The Mypy
documentation contains the following example:
```python
def process(items: List[str]) -> None:
# 'items' has type List[str]
items = [item.split() for item in items]
# 'items' now has type List[List[str]]
...
```
This configuration change is especially helpful in reducing the number
of `# type: ignore` directives needed to handle the common pattern of:
* Accepting a filepath as a string
* Overwriting the variable using `filepath = ensure_path(filepath)`
These changes enable redefinition and remove all `# type: ignore`
directives rendered redundant by this change.
* 🏷 Add type annotation to converters mapping
* 🚨 Fix Mypy error in convert CLI argument verification
* 🏷 Improve type annotation on `resolve_dot_names` helper
* 🏷 Add type annotations for `Vocab` attributes `strings` and `vectors`
* 🏷 Add type annotations for more `Vocab` attributes
* 🏷 Add loose type annotation for gold data compilation
* 🏷 Improve `_format_labels` type annotation
* 🏷 Fix `get_lang_class` type annotation
* 🏷 Loosen return type of `Language.evaluate`
* 🏷 Don't accept `Scorer` in `handle_scores_per_type`
* 🏷 Add `string_to_list` overloads
* 🏷 Fix non-Optional command-line options
* 🙈 Ignore redefinition of `wandb_logger` in `loggers.py`
* ➕ Install `typing_extensions` in Python 3.8+
The `typing_extensions` package states that it should be used when
"writing code that must be compatible with multiple Python versions".
Since SpaCy needs to support multiple Python versions, it should be used
when newer `typing` module members are required. One example of this is
`Literal`, which is available starting with Python 3.8.
Previously SpaCy tried to import `Literal` from `typing`, falling back
to `typing_extensions` if the import failed. However, Mypy doesn't seem
to be able to understand what `Literal` means when the initial import
means. Therefore, these changes modify how `compat` imports `Literal` by
always importing it from `typing_extensions`.
These changes also modify how `typing_extensions` is installed, so that
it is a requirement for all Python versions, including those greater
than or equal to 3.8.
* 🏷 Improve type annotation for `Language.pipe`
These changes add a missing overload variant to the type signature of
`Language.pipe`. Additionally, the type signature is enhanced to allow
type checkers to differentiate between the two overload variants based
on the `as_tuple` parameter.
Fixes#8772
* ➖ Don't install `typing-extensions` in Python 3.8+
After more detailed analysis of how to implement Python version-specific
type annotations using SpaCy, it has been determined that by branching
on a comparison against `sys.version_info` can be statically analyzed by
Mypy well enough to enable us to conditionally use
`typing_extensions.Literal`. This means that we no longer need to
install `typing_extensions` for Python versions greater than or equal to
3.8! 🎉
These changes revert previous changes installing `typing-extensions`
regardless of Python version and modify how we import the `Literal` type
to ensure that Mypy treats it properly.
* resolve mypy errors for Strict pydantic types
* refactor code to avoid missing return statement
* fix types of convert CLI command
* avoid list-set confustion in debug_data
* fix typo and formatting
* small fixes to avoid type ignores
* fix types in profile CLI command and make it more efficient
* type fixes in projects CLI
* put one ignore back
* type fixes for render
* fix render types - the sequel
* fix BaseDefault in language definitions
* fix type of noun_chunks iterator - yields tuple instead of span
* fix types in language-specific modules
* 🏷 Expand accepted inputs of `get_string_id`
`get_string_id` accepts either a string (in which case it returns its
ID) or an ID (in which case it immediately returns the ID). These
changes extend the type annotation of `get_string_id` to indicate that
it can accept either strings or IDs.
* 🏷 Handle override types in `combine_score_weights`
The `combine_score_weights` function allows users to pass an `overrides`
mapping to override data extracted from the `weights` argument. Since it
allows `Optional` dictionary values, the return value may also include
`Optional` dictionary values.
These changes update the type annotations for `combine_score_weights` to
reflect this fact.
* 🏷 Fix tokenizer serialization method signatures in `DummyTokenizer`
* 🏷 Fix redefinition of `wandb_logger`
These changes fix the redefinition of `wandb_logger` by giving a
separate name to each `WandbLogger` version. For
backwards-compatibility, `spacy.train` still exports `wandb_logger_v3`
as `wandb_logger` for now.
* more fixes for typing in language
* type fixes in model definitions
* 🏷 Annotate `_RandomWords.probs` as `NDArray`
* 🏷 Annotate `tok2vec` layers to help Mypy
* 🐛 Fix `_RandomWords.probs` type annotations for Python 3.6
Also remove an import that I forgot to move to the top of the module 😅
* more fixes for matchers and other pipeline components
* quick fix for entity linker
* fixing types for spancat, textcat, etc
* bugfix for tok2vec
* type annotations for scorer
* add runtime_checkable for Protocol
* type and import fixes in tests
* mypy fixes for training utilities
* few fixes in util
* fix import
* 🐵 Remove unused `# type: ignore` directives
* 🏷 Annotate `Language._components`
* 🏷 Annotate `spacy.pipeline.Pipe`
* add doc as property to span.pyi
* small fixes and cleanup
* explicit type annotations instead of via comment
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com>
Co-authored-by: svlandeg <svlandeg@github.com>
* Update universe plugins
* Adjust azure trigger
* Add init to tests/universe
* deliberatly trying to break the universe to see if the CI catches it
* revert
Co-authored-by: svlandeg <svlandeg@github.com>
* Update universe plugins
* Adjust azure trigger
* Add init to tests/universe
* deliberatly trying to break the universe to see if the CI catches it
* revert
Co-authored-by: svlandeg <svlandeg@github.com>
* remove text argument from W108 to enable 'once' filtering
* include the option of partial POS annotation
* fix typo
* Update spacy/errors.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Add test for case where parser overwrite annotations
* Move test to its own file
Also add note about how other tokens modify results.
* Fix xfail decorator
* Add util function to unique lists and preserve order
* Use unique function instead of list(set())
list(set()) has the issue that it's not consistent between runs of the
Python interpreter, so order can vary.
list(set()) calls were left in a few places where they were behind calls
to sorted(). I think in this case the calls to list() can be removed,
but this commit doesn't do that.
* Use the existing pattern for this
* Fix inconsistency
This makes the failing test pass, so that behavior is consistent whether
patterns are added in one call or two.
The issue is that the hash for patterns depended on the index of the
pattern in the list of current patterns, not the list of total patterns,
so a second call would get identical match ids.
* Add illustrative test case
* Add failing test for remove case
Patterns are not removed from the internal matcher on calls to remove,
which causes spurious weird matches (or misses).
* Fix removal issue
Remove patterns from the internal matcher.
* Check that the single add call also gets no matches
Hyphen is unsuitable because of interactions with the JA data fields,
but pipe is also unsuitable because it has a different meaning in UD
data, so it's better to use something that has no significance in either
case. So this uses semicolon.
* use language-matching to allow language code aliases
Signed-off-by: Elia Robyn Speer <elia@explosion.ai>
* link to "IETF language tags" in docs
Signed-off-by: Elia Robyn Speer <elia@explosion.ai>
* Make requirements consistent
Signed-off-by: Elia Robyn Speer <elia@explosion.ai>
* change "two-letter language ID" to "IETF language tag" in language docs
Signed-off-by: Elia Robyn Speer <elia@explosion.ai>
* use langcodes 3.2 and handle language-tag errors better
Signed-off-by: Elia Robyn Speer <elia@explosion.ai>
* all unknown language codes are ImportErrors
Signed-off-by: Elia Robyn Speer <elia@explosion.ai>
Co-authored-by: Elia Robyn Speer <elia@explosion.ai>
Since a component may reference anything in the vocab, share the full
vocab when loading source components and vectors (which will include
`strings` as of #8909).
When loading a source component from a config, save and restore the
vocab state after loading source pipelines, in particular to preserve
the original state without vectors, since `[initialize.vectors]
= null` skips rather than resets the vectors.
The vocab references are not synced for components loaded with
`Language.add_pipe(source=)` because the pipelines are already loaded
and not necessarily with the same vocab. A warning could be added in
`Language.create_pipe_from_source` that it may be necessary to save and
reload before training, but it's a rare enough case that this kind of
warning may be too noisy overall.
* Use morph for extra Japanese tokenizer info
Previously Japanese tokenizer info that didn't correspond to Token
fields was put in user data. Since spaCy core should avoid touching user
data, this moves most information to the Token.morph attribute. It also
adds the normalized form, which wasn't exposed before.
The subtokens, which are a list of full tokens, are still added to user
data, except with the default tokenizer granualarity. With the default
tokenizer settings the subtokens are all None, so in this case the user
data is simply not set.
* Update tests
Also adds a new test for norm data.
* Update docs
* Add Japanese morphologizer factory
Set the default to `extend=True` so that the morphologizer does not
clobber the values set by the tokenizer.
* Use the norm_ field for normalized forms
Before this commit, normalized forms were put in the "norm" field in the
morph attributes. I am not sure why I did that instead of using the
token morph, I think I just forgot about it.
* Skip test if sudachipy is not installed
* Fix import
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Add link to Discussions FAQ
* Remove old FAQ entries
I think these are no longer relevant.
- no-cache-dir: affected pip versions are *very* old now
- narrow unicode: not an issue from py3.3+
- utf-8 osx: upstream bug closed in 2019
Some of the other issues are also maybe not frequent.
* Add link to Discussions FAQ
* Remove old FAQ entries
I think these are no longer relevant.
- no-cache-dir: affected pip versions are *very* old now
- narrow unicode: not an issue from py3.3+
- utf-8 osx: upstream bug closed in 2019
Some of the other issues are also maybe not frequent.
* Add overwrite settings for more components
For pipeline components where it's relevant and not already implemented,
add an explicit `overwrite` setting that controls whether
`set_annotations` overwrites existing annotation.
For the `morphologizer`, add an additional setting `extend`, which
controls whether the existing features are preserved.
* +overwrite, +extend: overwrite values of existing features, add any new
features
* +overwrite, -extend: overwrite completely, removing any existing
features
* -overwrite, +extend: keep values of existing features, add any new
features
* -overwrite, -extend: do not modify the existing value if set
In all cases an unset value will be set by `set_annotations`.
Preserve current overwrite defaults:
* True: morphologizer, entity linker
* False: tagger, sentencizer, senter
* Add backwards compat overwrite settings
* Put empty line back
Removed by accident in last commit
* Set backwards-compatible defaults in __init__
Because the `TrainablePipe` serialization methods update `cfg`, there's
no straightforward way to detect whether models serialized with a
previous version are missing the overwrite settings.
It would be possible in the sentencizer due to its separate
serialization methods, however to keep the changes parallel, this also
sets the default in `__init__`.
* Remove traces
Co-authored-by: Paul O'Leary McCann <polm@dampfkraft.com>
* factor out the WandB logger into spacy-loggers
Signed-off-by: Elia Robyn Speer <gh@arborelia.net>
* depend on spacy-loggers so they are available
Signed-off-by: Elia Robyn Speer <gh@arborelia.net>
* remove docs of spacy.WandbLogger.v2 (moved to spacy-loggers)
Signed-off-by: Elia Robyn Speer <elia@explosion.ai>
* Version number suggestions from code review
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* update references to WandbLogger
Signed-off-by: Elia Robyn Speer <elia@explosion.ai>
* make order of deps more consistent
Signed-off-by: Elia Robyn Speer <elia@explosion.ai>
Co-authored-by: Elia Robyn Speer <elia@explosion.ai>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update Makefile
For more recent python version
* updated for bsc changes
New tokenization changes
* Update test_text.py
* updating tests and requirements
* changed failed test in test/lang/ca
changed failed test in test/lang/ca
* Update .gitignore
deleted stashed changes line
* back to python 3.6 and remove transformer requirements
As per request
* Update test_exception.py
Change the test
* Update test_exception.py
Remove test print
* Update Makefile
For more recent python version
* updated for bsc changes
New tokenization changes
* updating tests and requirements
* Update requirements.txt
Removed spacy-transfromers from requirements
* Update test_exception.py
Added final punctuation to ensure consistency
* Update Makefile
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Format
* Update test to check all tokens
Co-authored-by: cayorodriguez <crodriguezp@gmail.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Accept Doc input in pipelines
Allow `Doc` input to `Language.__call__` and `Language.pipe`, which
skips `Language.make_doc` and passes the doc directly to the pipeline.
* ensure_doc helper function
* avoid running multiple processes on GPU
* Update spacy/tests/test_language.py
Co-authored-by: svlandeg <svlandeg@github.com>
* Validate pos values when creating Doc
* Add clear error when setting invalid pos
This also changes the error language slightly.
* Fix variable name
* Update spacy/tokens/doc.pyx
* Test that setting invalid pos raises an error
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* First take at StringStore/Vocab docs
Things to check:
1. The mysterious vocab members
2. How to make table of contents? Is it autogenerated?
3. Anything I missed / needs more detail?
* Update docs
* Apply suggestions from code review
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Updates based on review feedback
* Minor fix
* Move example code down
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Remove two attributes marked for removal in 3.1
* Add back unused ints with changed names
* Change data_dir to _unused_object
This is still kept in the type definition, but I removed it from the
serialization code.
* Put serialization code back for now
Not sure how this interacts with old serialized models yet.
* Replace all basestring references with unicode
`basestring` was a compatability type introduced by Cython to make
dealing with utf-8 strings in Python2 easier. In Python3 it is
equivalent to the unicode (or str) type.
I replaced all references to basestring with unicode, since that was
used elsewhere, but we could also just replace them with str, which
shoudl also be equivalent.
All tests pass locally.
* Replace all references to unicode type with str
Since we only support python3 this is simpler.
* Remove all references to unicode type
This removes all references to the unicode type across the codebase and
replaces them with `str`, which makes it more drastic than the prior
commits. In order to make this work importing `unicode_literals` had to
be removed, and one explicit unicode literal also had to be removed (it
is unclear why this is necessary in Cython with language level 3, but
without doing it there were errors about implicit conversion).
When `unicode` is used as a type in comments it was also edited to be
`str`.
Additionally `coding: utf8` headers were removed from a few files.
* Handle spacy-legacy in package CLI for dependencies
* Implement legacy backoff in spacy registry.find
* Remove unused import
* Update and format test
* pass alignments to callbacks
* refactor for single callback loop
* Update spacy/matcher/matcher.pyx
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Fix surprises when asking for the root of a git repo
In the case of the first asset I wanted to get from git, the data I
wanted was the entire repository. I tried leaving "path" blank, which
gave a less-than-helpful error, and then I tried `path: "/"`, which
started copying my entire filesystem into the project. The path I should
have used was "".
I've made two changes to make this smoother for others:
- The 'path' within a git clone defaults to ""
- If the path points outside of the tmpdir that the git clone goes
into, we fail with an error
Signed-off-by: Elia Robyn Speer <elia@explosion.ai>
* use a descriptive error instead of a default
plus some minor fixes from PR review
Signed-off-by: Elia Robyn Speer <elia@explosion.ai>
* check for None values in assets
Signed-off-by: Elia Robyn Speer <elia@explosion.ai>
Co-authored-by: Elia Robyn Speer <elia@explosion.ai>
* Add textcat docs
* Add NER docs
* Add Entity Linker docs
* Add assigned fields docs for the tagger
This also adds a preamble, since there wasn't one.
* Add morphologizer docs
* Add dependency parser docs
* Update entityrecognizer docs
This is a little weird because `Doc.ents` is the only thing assigned to,
but it's actually a bidirectional property.
* Add token fields for entityrecognizer
* Fix section name
* Add entity ruler docs
* Add lemmatizer docs
* Add sentencizer/recognizer docs
* Update website/docs/api/entityrecognizer.md
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update website/docs/api/entityruler.md
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update website/docs/api/tagger.md
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update website/docs/api/entityruler.md
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update type for Doc.ents
This was `Tuple[Span, ...]` everywhere but `Tuple[Span]` seems to be
correct.
* Run prettier
* Apply suggestions from code review
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Run prettier
* Add transformers section
This basically just moves and renames the "custom attributes" section
from the bottom of the page to be consistent with "assigned attributes"
on other pages.
I looked at moving the paragraph just above the section into the
section, but it includes the unrelated registry additions, so it seemed
better to leave it unchanged.
* Make table header consistent
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Add textcat docs
* Add NER docs
* Add Entity Linker docs
* Add assigned fields docs for the tagger
This also adds a preamble, since there wasn't one.
* Add morphologizer docs
* Add dependency parser docs
* Update entityrecognizer docs
This is a little weird because `Doc.ents` is the only thing assigned to,
but it's actually a bidirectional property.
* Add token fields for entityrecognizer
* Fix section name
* Add entity ruler docs
* Add lemmatizer docs
* Add sentencizer/recognizer docs
* Update website/docs/api/entityrecognizer.md
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update website/docs/api/entityruler.md
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update website/docs/api/tagger.md
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update website/docs/api/entityruler.md
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update type for Doc.ents
This was `Tuple[Span, ...]` everywhere but `Tuple[Span]` seems to be
correct.
* Run prettier
* Apply suggestions from code review
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Run prettier
* Add transformers section
This basically just moves and renames the "custom attributes" section
from the bottom of the page to be consistent with "assigned attributes"
on other pages.
I looked at moving the paragraph just above the section into the
section, but it includes the unrelated registry additions, so it seemed
better to leave it unchanged.
* Make table header consistent
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Fix inference of epoch_resume
When an epoch_resume value is not specified individually, it can often
be inferred from the filename. The value inference code was there but
the value wasn't passed back to the training loop.
This also adds a specific error in the case where no epoch_resume value
is provided and it can't be inferred from the filename.
* Add new error
* Always use the epoch resume value if specified
Before this the value in the filename was used if found
* Start Listeners documentation
* intro tabel of different architectures
* initialization, linking, dim inference
* internal comm (WIP)
* expand internal comm section
* frozen components and replacing listeners
* various small fixes
* fix content table
* fix link
* overfitting test on non-overlapping entities
* add failing overfitting test for overlapping entities
* failing test for list comprehension
* remove test that was put in separate PR
* bugfix
* cleanup
* test for error after Doc has been garbage collected
* warn about using a SpanGroup when the Doc has been garbage collected
* add warning to the docs
* rephrase slightly
* raise error instead of warning
* update
* move warning to doc property
* Fix incorrect pickling of Japanese and Korean pipelines, which led to
the entire pipeline being reset if pickled
* Enable pickling of Vietnamese tokenizer
* Update tokenizer APIs for Chinese, Japanese, Korean, Thai, and
Vietnamese so that only the `Vocab` is required for initialization
* Refactor to use list comps and enumerate.
Replace loops that append to a list with a list comprehensions where this does not change the behavior; replace range(len(...)) loops with enumerate. Correct one typo in a comment. Replace a call to set() with a set literal.
* Undo double assignment.
Expand `tokens_to_key[j] = k = self._get_matcher_key(key, i, j)` to two statements.
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Sign contributors agreement
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Add training data section
Not entirely sure this is in the right location on the page - maybe it
should be after quickstart?
* Add pointer from binary format to training data section
* Minor cleanup
* Add to ToC, fix filename
* Update website/docs/usage/training.md
Co-authored-by: Ines Montani <ines@ines.io>
* Update website/docs/usage/training.md
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update website/docs/usage/training.md
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Move the training data section further down the page
* Update website/docs/usage/training.md
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update website/docs/usage/training.md
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Run prettier
Co-authored-by: Ines Montani <ines@ines.io>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Add training data section
Not entirely sure this is in the right location on the page - maybe it
should be after quickstart?
* Add pointer from binary format to training data section
* Minor cleanup
* Add to ToC, fix filename
* Update website/docs/usage/training.md
Co-authored-by: Ines Montani <ines@ines.io>
* Update website/docs/usage/training.md
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update website/docs/usage/training.md
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Move the training data section further down the page
* Update website/docs/usage/training.md
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update website/docs/usage/training.md
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Run prettier
Co-authored-by: Ines Montani <ines@ines.io>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Allow passing in array vars for speedup
This fixes#8845. Not sure about the docstring changes here...
* Update docs
Types maybe need more detail? Maybe not?
* Run prettier on docs
* Update spacy/tokens/span.pyx
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Add scorer option to components
Add an optional `scorer` parameter to all pipeline components. If a
scoring function is provided, it overrides the default scoring method
for that component.
* Add registered scorers for all components
* Add `scorers` registry
* Move all scoring methods outside of components as independent
functions and register
* Use the registered scoring methods as defaults in configs and inits
Additional:
* The scoring methods no longer have access to the full component, so
use settings from `cfg` as default scorer options to handle settings
such as `labels`, `threshold`, and `positive_label`
* The `attribute_ruler` scoring method no longer has access to the
patterns, so all scoring methods are called
* Bug fix: `spancat` scoring method is updated to set `allow_overlap` to
score overlapping spans correctly
* Update Russian lemmatizer to use direct score method
* Check type of cfg in Pipe.score
* Fix check
* Update spacy/pipeline/sentencizer.pyx
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Remove validate_examples from scoring functions
* Use Pipe.labels instead of Pipe.cfg["labels"]
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Add missing punctuation for Tigrinya and Amharic
* Fix numeral and ordinal numbers for Tigrinya
- Amharic was used in many cases
- Also fixed some typos
* Update Tigrinya stop-words
* Contributor agreement for fgaim
* Fix typo in "ti" lang test
* Remove multi-word entries from numbers and ordinals
* Add scores to output in spancat
This exposes the scores as an attribute on the SpanGroup. Includes a
basic test.
* Add basic doc note
* Vectorize score calcs
* Add "annotation format" section
* Update website/docs/api/spancategorizer.md
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Clean up doc section
* Ran prettier on docs
* Get arrays off the gpu before iterating over them
* Remove int() calls
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Add more stop words and Improve the readability
* Add and categorize the tokenizer exceptions for `bg` lang
* Create syrull.md
* Add references for the additional stop words and tokenizer exc abbrs
* Add stub files for main API classes
* Add contributor agreement for ezorita
* Update types for ndarray and hash()
* Fix __getitem__ and __iter__
* Add attributes of Doc and Token classes
* Overload type hints for Span.__getitem__
* Fix type hint overload for Span.__getitem__
Co-authored-by: Luca Dorigo <dorigoluca@gmail.com>
* Fix check for RIGHT_ATTRs in dep matcher
If a non-anchor node does not have RIGHT_ATTRS, the dep matcher throws
an E100, which says that non-anchor nodes must have LEFT_ID, REL_OP, and
RIGHT_ID. It specifically does not say RIGHT_ATTRS is required.
A blank RIGHT_ATTRS is also valid, and patterns with one will be
excepted. While not normal, sometimes a REL_OP is enough to specify a
non-anchor node - maybe you just want the head of another node
unconditionally, for example.
This change just sets RIGHT_ATTRS to {} if not present. Alternatively
changing E100 to state RIGHT_ATTRS is required could also be reasonable.
* Fix test
This test was written on the assumption that if `RIGHT_ATTRS` isn't
present an error will be raised. Since the proposed changes make it so
an error won't be raised this is no longer necessary.
* Revert test, update error message
Error message now lists missing keys, and RIGHT_ATTRS is required.
* Use list of required keys in error message
Also removes unused key param arg.
* Pass excludes when serializing vocab
Additional minor bug fix:
* Deserialize vocab in `EntityLinker.from_disk`
* Add test for excluding strings on load
* Fix formatting
* Support list values and IS_INTERSECT in Matcher
* Support list values as token attributes for set operators, not just as
pattern values.
* Add `IS_INTERSECT` operator.
* Fix incorrect `ISSUBSET` and `ISSUPERSET` in schema and docs.
* Rename IS_INTERSECT to INTERSECTS
* Support list values and IS_INTERSECT in Matcher
* Support list values as token attributes for set operators, not just as
pattern values.
* Add `IS_INTERSECT` operator.
* Fix incorrect `ISSUBSET` and `ISSUPERSET` in schema and docs.
* Rename IS_INTERSECT to INTERSECTS
* Add ancient Greek language support
Initial commit
* Contributor Agreement
* grc tokenizer test added and files formatted with black, unnecessary import removed
Co-Authored-By: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Commas in lists fixed. __init__py added to test
* Update lex_attrs.py
* Update stop_words.py
* Update stop_words.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* ✨ implement noun_chunks for dutch language
* copy/paste FR and SV syntax iterators to accomodate UD tags
* added tests with dutch text
* signed contributor agreement
* 🐛 fix noun chunks generator
* built from scratch
* define noun chunk as a single Noun-Phrase
* includes some corner cases debugging (incorrect POS tagging)
* test with provided annotated sample (POS, DEP)
* ✅ fix failing test
* CI pipeline did not like the added sample file
* add the sample as a pytest fixture
* Update spacy/lang/nl/syntax_iterators.py
* Update spacy/lang/nl/syntax_iterators.py
Code readability
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update spacy/tests/lang/nl/test_noun_chunks.py
correct comment
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* finalize code
* change "if next_word" into "if next_word is not None"
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
The autoblack job is an occasional cleanup job. If it runs on forks and
those PRs are accepted the git history will be weird and that doesn't
help anyone.
The way to make the job not run on forks is a little non-obvious but
based on this thread.
https://github.com/prisma/prisma/issues/3539
* avoid msg var impliciteness
* rename local msg
* Add CI tests for debug data and train
* Adjust debug data CLI test
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Add the right return type for Language.pipe and an overload for the as_tuples version
* Reformat, tidy up
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Fix vectors check for sourced components
Since vectors are not loaded when components are sourced, store a hash
for the vectors of each sourced component and compare it to the loaded
vectors after the vectors are loaded from the `[initialize]` block.
* Pop temporary info
* Remove stored hash in remove_pipe
* Add default for pop
* Add additional convert/debug/assemble CLI tests
* Raise an error for textcat with <2 labels
Raise an error if initializing a `textcat` component without at least
two labels.
* Add similar note to docs
* Update positive_label description in API docs
This uses some new features related to Issue Templates to help direct
more people to Discussions.
1. Change the Discussions option to link to Discussions
2. Add a link to the FAQ
3. Disable blank issues
* Draft spancat model
* Add spancat model
* Add test for extract_spans
* Add extract_spans layer
* Upd extract_spans
* Add spancat model
* Add test for spancat model
* Upd spancat model
* Update spancat component
* Upd spancat
* Update spancat model
* Add quick spancat test
* Import SpanCategorizer
* Fix SpanCategorizer component
* Import SpanGroup
* Fix span extraction
* Fix import
* Fix import
* Upd model
* Update spancat models
* Add scoring, update defaults
* Update and add docs
* Fix type
* Update spacy/ml/extract_spans.py
* Auto-format and fix import
* Fix comment
* Fix type
* Fix type
* Update website/docs/api/spancategorizer.md
* Fix comment
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Better defense
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Fix labels list
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update spacy/ml/extract_spans.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Update spacy/pipeline/spancat.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Set annotations during update
* Set annotations in spancat
* fix imports in test
* Update spacy/pipeline/spancat.py
* replace MaxoutLogistic with LinearLogistic
* fix config
* various small fixes
* remove set_annotations parameter in update
* use our beloved tupley format with recent support for doc.spans
* bugfix to allow renaming the default span_key (scores weren't showing up)
* use different key in docs example
* change defaults to better-working parameters from project (WIP)
* register spacy.extract_spans.v1 for legacy purposes
* Upd dev version so can build wheel
* layers instead of architectures for smaller building blocks
* Update website/docs/api/spancategorizer.md
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update website/docs/api/spancategorizer.md
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Include additional scores from overrides in combined score weights
* Parameterize spans key in scoring
Parameterize the `SpanCategorizer` `spans_key` for scoring purposes so
that it's possible to evaluate multiple `spancat` components in the same
pipeline.
* Use the (intentionally very short) default spans key `sc` in the
`SpanCategorizer`
* Adjust the default score weights to include the default key
* Adjust the scorer to use `spans_{spans_key}` as the prefix for the
returned score
* Revert addition of `attr_name` argument to `score_spans` and adjust
the key in the `getter` instead.
Note that for `spancat` components with a custom `span_key`, the score
weights currently need to be modified manually in
`[training.score_weights]` for them to be available during training. To
suppress the default score weights `spans_sc_p/r/f` during training, set
them to `null` in `[training.score_weights]`.
* Update website/docs/api/scorer.md
* Fix scorer for spans key containing underscore
* Increment version
* Add Spans to Evaluate CLI (#8439)
* Add Spans to Evaluate CLI
* Change to spans_key
* Add spans per_type output
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Fix spancat GPU issues (#8455)
* Fix GPU issues
* Require thinc >=8.0.6
* Switch to glorot_uniform_init
* Fix and test ngram suggester
* Include final ngram in doc for all sizes
* Fix ngrams for docs of the same length as ngram size
* Handle batches of docs that result in no ngrams
* Add tests
Co-authored-by: Ines Montani <ines@ines.io>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Co-authored-by: Nirant <NirantK@users.noreply.github.com>
* Use minor version for compatibility check
* Use minor version of compatibility table
* Soften warning message about incompatible models
* Add test for presence of current version in compatibility table
* Add test for download compatibility table
* Use minor version of lower pin in error message if possible
* Fall back to spacy_git_version if available
* Fix unknown version string
* Don't use the same vocab for source models
The source models should not be loaded with the vocab from the current
pipeline because this loads the vectors from the source model into the
current vocab.
The strings are all copied in `Language.create_pipe_from_source`, so if
the vectors are configured correctly in the current pipeline, the
sourced component will work as expected. If there is a vector mismatch,
a warning is shown. (It's not possible to inspect whether the vectors
are actually used by the component, so a warning is the best option.)
* Update comment on source model loading
* Copy rather than move files to top-level of package
* Add all files to `MANIFEST.in` (primarily for older versions of pip)
* Include the `README.md` contents as `long_description` in the setup
* Support a cfg field in transition system
* Make NER 'has gold' check use right alignment for span
* Pass 'negative_samples_key' property into NER transition system
* Add field for negative samples to NER transition system
* Check neg_key in NER has_gold
* Support negative examples in NER oracle
* Test for negative examples in NER
* Fix name of config variable in NER
* Remove vestiges of old-style partial annotation
* Remove obsolete tests
* Add comment noting lack of support for negative samples in parser
* Additions to "neg examples" PR (#8201)
* add custom error and test for deprecated format
* add test for unlearning an entity
* add break also for Begin's cost
* add negative_samples_key property on Parser
* rename
* extend docs & fix some older docs issues
* add subclass constructors, clean up tests, fix docs
* add flaky test with ValueError if gold parse was not found
* remove ValueError if n_gold == 0
* fix docstring
* Hack in environment variables to try out training
* Remove hack
* Remove NER hack, and support 'negative O' samples
* Fix O oracle
* Fix transition parser
* Remove 'not O' from oracle
* Fix NER oracle
* check for spans in both gold.ents and gold.spans and raise if so, to prevent memory access violation
* use set instead of list in consistency check
Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* implement textcat resizing for TextCatCNN
* resizing textcat in-place
* simplify code
* ensure predictions for old textcat labels remain the same after resizing (WIP)
* fix for softmax
* store softmax as attr
* fix ensemble weight copy and cleanup
* restructure slightly
* adjust documentation, update tests and quickstart templates to use latest versions
* extend unit test slightly
* revert unnecessary edits
* fix typo
* ensemble architecture won't be resizable for now
* use resizable layer (WIP)
* revert using resizable layer
* resizable container while avoid shape inference trouble
* cleanup
* ensure model continues training after resizing
* use fill_b parameter
* use fill_defaults
* resize_layer callback
* format
* bump thinc to 8.0.4
* bump spacy-legacy to 3.0.6
* Added Italian POS-aware lemmatizer.
Also added the code used to build the lookup tables by POS.
* Create gtoffoli.md
* Add imports and format
* Remove helper script
* Use lemma_lookup instead of lemma_lookup_legacy
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
To avoid config errors during training when `[corpora.pretrain.path]` is
`None` with the default `spacy.JsonlCorpus.v1` reader, make the reader
path optional, similar to `spacy.Corpus.v1`.
* Change span lemmas to use original whitespace (fix#8368)
This is a redo of #8371 based off master.
The test for this required some changes to existing tests. I don't think
the changes were significant but I'd like someone to check them.
* Remove mystery docstring
This sentence was uncompleted for years, and now we will never know how
it ends.
* Fill in deps if not provided with heads
Before this change, if heads were passed without deps they would be
silently ignored, which could be confusing. See #8334.
* Use "dep" instead of a blank string
This is the customary placeholder dep. It might be better to show an
error here instead though.
* Throw error on heads without deps
* Add a test
* Fix tests
* Formatting
* Fix all tests
* Fix a test I missed
* Revise error message
* Clean up whitespace
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update cats score names in Scorer API docs
* Refer to performance in meta
* Update package naming/versions, lemmatizer details
* Minor formatting fixes
* Provide more explanation for cats_score_desc
* Provide language-specific lemmatizer defaults in API docs
Co-authored-by: Paul O'Leary McCann <polm@dampfkraft.com>
* Update Catalan language data
Update Catalan language data based on contributions from the Text Mining
Unit at the Barcelona Supercomputing Center:
https://github.com/TeMU-BSC/spacy4release/tree/main/lang_data
* Update tokenizer settings for UD Catalan AnCora
Update for UD Catalan AnCora v2.7 with merged multi-word tokens.
* Update test
* Move prefix patternt to more generic infix pattern
* Clean up
For the Russian and Ukrainian lemmatizers, restrict the `pymorphy2`
requirement to the mode `pymorphy2` so that lookup or other lemmatizer
modes can be loaded without installing `pymorphy2`.
* Update CI: update ubuntu image, add download test
* Switch instances to `ubuntu-18.04`
* Add model download test, currently only for one job with python 3.8
* Fix variable name
* Set variables explicitly
* Show warning if entity_ruler runs without patterns
* Show warning if matcher runs without patterns
* fix wording
* unit test for warning once (WIP)
* warn W036 only once
* cleanup
* create filter_warning helper
* Don't add duplicate patterns (fix#8216)
* Refactor EntityRuler init
This simplifies the EntityRuler init code. This is helpful as prep for
allowing the EntityRuler to reset itself.
* Make EntityRuler.clear reset matchers
Includes a new test for this.
* Tidy PhraseMatcher instantiation
Since the attr can be None safely now, the guard if is no longer
required here.
Also renamed the `_validate` attr. Maybe it's not needed?
* Fix NER test
* Add test to make sure patterns aren't increasing
* Move test to regression tests
* "y" etc.
Many changes described in pull request
* Update spacy/lang/fr/stop_words.py
* Update spacy/lang/fr/stop_words.py
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
The attributes `PROB`, `CLUSTER` and `SENT_END` are not supported by
`Lexeme.get_struct_attr` so should not be included through `attrs.IDS`
as supported attributes in `Doc.to_array` and other methods.
* Show warning if entity_ruler runs without patterns
* Show warning if matcher runs without patterns
* fix wording
* unit test for warning once (WIP)
* warn W036 only once
* cleanup
* create filter_warning helper
* Add all symbols in Unicode Currency Symbols block
In #8102 it came up that the rupee symbol was treated different from
dollar / euro / yen symbols. This adds many symbols not already
included.
* Fix test
* Fix training test
The behavior of `spacy.Corpus.v1` is unexpected enough for `max_length
!= 0` that `0` is a better default for users creating a new config with
the quickstart.
If not, documents are skipped, sometimes the entire corpus is skipped,
and sometimes documents are (quite unexpectedly for your average user)
split into sentences.
* unit test for pickling KB
* add pickling test for NEL
* KB to_bytes and from_bytes
* NEL to_bytes and from_bytes
* xfail pickle tests for now
* fix docs
* cleanup
* Minor updates to quickstart settings/instructions
* set default value of textcat exclusive to `false` until the default
checkbox behavior is updated
* add the `morphologizer` to the list of components
* add a note that v3.0.6+ is required
* Switch to warning above quickstart
* Undo changes to textcat default in quickstart
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Minor updates to quickstart settings/instructions
* set default value of textcat exclusive to `false` until the default
checkbox behavior is updated
* add the `morphologizer` to the list of components
* add a note that v3.0.6+ is required
* Switch to warning above quickstart
* Undo changes to textcat default in quickstart
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Fix range in Span.get_lca_matrix
Fix the adjusted token index / lca matrix index ranges for
`_get_lca_matrix` for spans.
* The range for `k` should correspond to the adjusted indices in
`lca_matrix` with the `start` indexed at `0`
* Update test for v3.x
* custom warning if the doc_bin is too large
* cleanup
* Update spacy/errors.py
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* fix numbering
* fixing numbering once more
* fixing this seems to be pretty hard
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Handle errors while multiprocessing
Handle errors while multiprocessing without hanging.
* Return the traceback for errors raised while processing a batch, which
can be handled by the top-level error handler
* Allow for shortened batches due to custom error handlers that ignore
errors and skip documents
* Define custom components at a higher level
* Also move up custom error handler
* Use simpler component for test
* Switch error type
* Adjust test
* Only call top-level error handler for exceptions
* Register custom test components within tests
Use global functions (so they can be pickled) but register the
components only within the individual tests.
* Check for unsupported cats values
* Only show labels if train/dev mismatched
* Don't show label counts (only counting positive labels seems odd)
* Use warnings for mismatched train/dev labels
* Adapt tokenization methods from `pyvi` to preserve text encoding and
whitespace
* Add serialization support similar to Chinese and Japanese
Note: as for Chinese and Japanese, some settings are duplicated in
`config.cfg` and `tokenizer/cfg`.
* Handle partial entities in Span.as_doc
In `Span.as_doc` replace partial entities at the beginning or end of the
span with missing entity annotation.
Fixes a bug where invalid entity annotation (no initial `B`) was
returned for an initial partial entity.
* Check for empty span in ents conversion
Note: `Span.as_doc()` will still fail on an empty span due to failures
in `Span.vector`.
* Preserve existing ENT_KB_ID annotation in NER
Preserve `ent_kb_id` annotation on existing entity spans, which is not
preserved by the transition system.
* Simplify kb_id assignment
* Simplify further
* Fix pretraining objectives fragment
The fragment here is reused from a heading higher up, so you couldn't
link to this section.
* Fix section link to new fragment
This came up in #7878, but if --resume-path is a directory then loading
the weights will fail. On Linux this will give a straightforward error
message, but on Windows it gives "Permission Denied", which is
confusing.
* Fix percent unk display
This was showing (ratio %), so 10% would show as 0.10%. Fix by
multiplying ration by 100.
Might want to add a warning if this is over a threshold.
* Only show whole-integer percents
* Add training option to set annotations on update
Add a `[training]` option called `set_annotations_on_update` to specify
a list of components for which the predicted annotations should be set
on `example.predicted` immediately after that component has been
updated. The predicted annotations can be accessed by later components
in the pipeline during the processing of the batch in the same `update`
call.
* Rename to annotates / annotating_components
* Add test for `annotating_components` when training from config
* Add documentation
* Add empty lines at the end of Python files
* Only prepend the lang code if it's not there already
* Update spacy/cli/package.py
* fix whitespace stripping
* Set up CI for tests with GPU agent
* Update tests for enabled GPU
* Fix steps filename
* Add parallel build jobs as a setting
* Fix test requirements
* Fix install test requirements condition
* Fix pipeline models test
* Reset current ops in prefer/require testing
* Fix more tests
* Remove separate test_models test
* Fix regression 5551
* fix StaticVectors for GPU use
* fix vocab tests
* Fix regression test 5082
* Move azure steps to .github and reenable default pool jobs
* Consolidate/rename azure steps
Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com>
* Add callback to copy vocab/tokenizer from model
Add callback `spacy.copy_from_base_model.v1` to copy the tokenizer
settings and/or vocab (including vectors) from a base model.
* Move spacy.copy_from_base_model.v1 to spacy.training.callbacks
* Add documentation
* Modify to specify model as tokenizer and vocab params
* Update sent_starts in Example.from_dict
Update `sent_starts` for `Example.from_dict` so that `Optional[bool]`
values have the same meaning as for `Token.is_sent_start`.
Use `Optional[bool]` as the type for sent start values in the docs.
* Use helper function for conversion to ternary ints
* Fix tokenizer cache flushing
Fix/simplify tokenizer init detection in order to fix cache flushing
when properties are modified.
* Remove init reloading logic
* Remove logic disabling `_reload_special_cases` on init
* Setting `rules` last in `__init__` (as before) means that setting
other properties doesn't reload any special cases
* Reset `rules` first in `from_bytes` so that setting other properties
during deserialization doesn't reload any special cases
unnecessarily
* Reset all properties in `Tokenizer.from_bytes` to allow any settings
to be `None`
* Also reset special matcher when special cache is flushed
* Remove duplicate special case validation
* Add test for special cases flushing
* Extend test for tokenizer deserialization of None values
* Replace negative rows with 0 in StaticVectors
Replace negative row indices with 0-vectors in `StaticVectors`.
* Increase versions related to StaticVectors
* Increase versions of all architctures and layers related to
`StaticVectors`
* Improve efficiency of 0-vector operations
Parallel `spacy-legacy` PR: https://github.com/explosion/spacy-legacy/pull/5
* Update config defaults to new versions
* Update docs
* Update processing-pipelines.md
Under "things to try," inform users they can save metadata when using nlp.pipe(foobar, as_tuples=True)
Link to a new example on the attributes page detailing the following:
> ```
> data = [
> ("Some text to process", {"meta": "foo"}),
> ("And more text...", {"meta": "bar"})
> ]
>
> for doc, context in nlp.pipe(data, as_tuples=True):
> # Let's assume you have a "meta" extension registered on the Doc
> doc._.meta = context["meta"]
> ```
from https://stackoverflow.com/questions/57058798/make-spacy-nlp-pipe-process-tuples-of-text-and-additional-information-to-add-as
* Updating the attributes section
Update the attributes section with example of how extensions can be used to store metadata.
* Update processing-pipelines.md
* Update processing-pipelines.md
Made as_tuples example executable and relocated to the end of the "Processing Text" section.
* Update processing-pipelines.md
* Update processing-pipelines.md
Removed extra line
* Reformat and rephrase
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update processing-pipelines.md
Under "things to try," inform users they can save metadata when using nlp.pipe(foobar, as_tuples=True)
Link to a new example on the attributes page detailing the following:
> ```
> data = [
> ("Some text to process", {"meta": "foo"}),
> ("And more text...", {"meta": "bar"})
> ]
>
> for doc, context in nlp.pipe(data, as_tuples=True):
> # Let's assume you have a "meta" extension registered on the Doc
> doc._.meta = context["meta"]
> ```
from https://stackoverflow.com/questions/57058798/make-spacy-nlp-pipe-process-tuples-of-text-and-additional-information-to-add-as
* Updating the attributes section
Update the attributes section with example of how extensions can be used to store metadata.
* Update processing-pipelines.md
* Update processing-pipelines.md
Made as_tuples example executable and relocated to the end of the "Processing Text" section.
* Update processing-pipelines.md
* Update processing-pipelines.md
Removed extra line
* Reformat and rephrase
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* Update Tokenizer.explain with special matches
Update `Tokenizer.explain` and the pseudo-code in the docs to include
the processing of special cases that contain affixes or whitespace.
* Handle optional settings in explain
* Add test for special matches in explain
Add test for `Tokenizer.explain` for special cases containing affixes.
* Set catalogue lower pin to v2.0.2
* Update importlib-metadata pins to match
* Require catalogue v2.0.3
Switch to vendored `importlib-metadata` v3.2.0 provided by `catalogue`.
* ensure vectors data is stored on right device
* ensure the added vector is on the right device
* move vector to numpy before iterating
* move best_rows to numpy before iterating
* Terminology: deprecated vs obsolete
Typically, deprecated is used for functionality that is bound to become unavailable but that can still be used. Obsolete is used for features that have been removed. In E941, I think what is meant is "obsolete" since loading a model by a shortcut simply does not work anymore (and throws an error). This is different from downloading a model with a shortcut, which is deprecated but still works.
In light of this, perhaps all other error codes should be checked as well.
* clarify that the link command is removed and not just deprecated
Co-authored-by: svlandeg <sofie.vanlandeghem@gmail.com>
* Update debug data further for v3
* Remove new/existing label distinction (new labels are not immediately
distinguishable because the pipeline is already initialized)
* Warn on missing labels in training data for all components except parser
* Separate textcat and textcat_multilabel sections
* Add section for morphologizer
* Reword missing label warnings
* Make vocab update in get_docs deterministic
The attribute `DocBin.strings` is a set. In `DocBin.get_docs`
a given vocab is updated by iterating over this set.
Iteration over a python set produces an arbitrary ordering,
therefore vocab is updated non-deterministically.
When training (fine-tuning) a spacy model, the base model's
vocabulary will be updated with the new vocabulary in the
training data in exactly the way described above. After
serialization, the file `model/vocab/strings.json` will
be sorted in an arbitrary way. This prevents reproducible
model training.
* Revert "Make vocab update in get_docs deterministic"
This reverts commit d6b87a2f558b52d66549b6a66c0af00e283ad628.
* Sort strings in StringStore serialization
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
* extend span scorer with consider_label and allow_overlap
* unit test for spans y2x overlap
* add score_spans unit test
* docs for new fields in scorer.score_spans
* rename to include_label
* spell out if-else for clarity
* rename to 'labeled'
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Data in the JSON format is split into sentences, and each sentence is
saved with is_sent_start flags. Currently the flags are 1 for the first
token and 0 for the others. When deserialized this results in a pattern
of True, None, None, None... which makes single-sentence documents look
as though they haven't had sentence boundaries set.
Since items saved in JSON format have been split into sentences already,
the is_sent_start values should all be True or False.
* Support match alignments
* change naming from match_alignments to with_alignments, add conditional flow if with_alignments is given, validate with_alignments, add related test case
* remove added errors, utilize bint type, cleanup whitespace
* fix no new line in end of file
* Minor formatting
* Skip alignments processing if as_spans is set
* Add with_alignments to Matcher API docs
* Update website/docs/api/matcher.md
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Support infinite generators for training corpora
Support a training corpus with an infinite generator in the `spacy
train` training loop:
* Revert `create_train_batches` to the state where an infinite generator
can be used as the in the first epoch of exactly one epoch without
resulting in a memory leak (`max_epochs != 1` will still result in a
memory leak)
* Move the shuffling for the first epoch into the corpus reader,
renaming it to `spacy.Corpus.v2`.
* Switch to training option for shuffling in memory
Training loop:
* Add option `training.shuffle_train_corpus_in_memory` that controls
whether the corpus is loaded in memory once and shuffled in the training
loop
* Revert changes to `create_train_batches` and rename to
`create_train_batches_with_shuffling` for use with `spacy.Corpus.v1` and
a corpus that should be loaded in memory
* Add `create_train_batches_without_shuffling` for a corpus that
should not be shuffled in the training loop: the corpus is merely
batched during training
Corpus readers:
* Restore `spacy.Corpus.v1`
* Add `spacy.ShuffledCorpus.v1` for a corpus shuffled in memory in the
reader instead of the training loop
* In combination with `shuffle_train_corpus_in_memory = False`, each
epoch could result in a different augmentation
* Refactor create_train_batches, validation
* Rename config setting to `training.shuffle_train_corpus`
* Refactor to use a single `create_train_batches` method with a
`shuffle` option
* Only validate `get_examples` in initialize step if:
* labels are required
* labels are not provided
* Switch back to max_epochs=-1 for streaming train corpus
* Use first 100 examples for stream train corpus init
* Always check validate_get_examples in initialize
* Adjust custom extension data when copying user data in `Span.as_doc()`
* Restrict `Doc.from_docs()` to adjusting offsets for custom extension
data
* Update test to use extension
* (Duplicate bug fix for character offset from #7497)
Merge data from `doc.spans` in `Doc.from_docs()`.
* Fix internal character offset set when merging empty docs (only
affects tokens and spans in `user_data` if an empty doc is in the list
of docs)
In the retokenizer, only reset sent starts (with
`set_children_from_head`) if the doc is parsed. If there is no parse,
merged tokens have the unset `token.is_sent_start == None` by default after
retokenization.
* Add util method for check
* Add new languages to list with lexeme norm tables
* Add check to all relevant components
* Add config details to warning message
Note that we're not actually inspecting the model config to see if
`NORM` is used as an attribute, so it may warn in cases where it's not
relevant.
See here:
https://github.com/explosion/spaCy/discussions/7463
Still need to check if there are any side effects of listeners being
present but not in the pipeline, but this commit will silence the
warnings.
* To allow default lookup lemmatization with a blank Russian model,
rename pymorphy2 lookup mode to `pymorphy2_lookup`
* Bug fix: update pymorphy2 lookup lemmatize to return list rather than
string
* add multi-label textcat to menu
* add infobox on textcat API
* add info to v3 migration guide
* small edits
* further fixes in doc strings
* add infobox to textcat architectures
* add textcat_multilabel to overview of built-in components
* spelling
* fix unrelated warn msg
* Add textcat_multilabel to quickstart [ci skip]
* remove separate documentation page for multilabel_textcategorizer
* small edits
* positive label clarification
* avoid duplicating information in self.cfg and fix textcat.score
* fix multilabel textcat too
* revert threshold to storage in cfg
* revert threshold stuff for multi-textcat
Co-authored-by: Ines Montani <ines@ines.io>
* Fix aborted/skipped augmentation for `spacy.orth_variants.v1` if
lowercasing was enabled for an example
* Simplify `spacy.orth_variants.v1` for `Example` vs. `GoldParse`
* Preserve reference tokenization in `spacy.lower_case.v1`
* initialize NLP with train corpus
* add more pretraining tests
* more tests
* function to fetch tok2vec layer for pretraining
* clarify parameter name
* test different objectives
* formatting
* fix check for static vectors when using vectors objective
* clarify docs
* logger statement
* fix init_tok2vec and proc.initialize order
* test training after pretraining
* add init_config tests for pretraining
* pop pretraining block to avoid config validation errors
* custom errors
* Fix patience for identical scores
Fix training patience so that the earliest best step is chosen for
identical max scores.
* Restore break, remove print
* Explicitly define best_step for clarity
* Add hint for --gpu-id to CLI device info
If the user has `cupy` and an available GPU, add a hint about using
`--gpu-id 0` to the CLI output.
* Undo change to original CPU message
* Fix `is_cython_func` for imported code loaded under `python_code`
module name
* Add `make_named_tempfile` context manager to test utils to test
loading of imported code
* Add test for validation of `initialize` params in custom module
Set the `include_dirs` in each `Extension` rather than in `setup()` to
handle the case where there is a custom `distutils.cfg` that modifies
the include paths, in particular for python from homebrew.
Fix class variable and init for `UkrainianLemmatizer` so that it loads
the `uk` dictionaries rather than having the parent `RussianLemmatizer`
override with the `ru` settings.
Now that `nlp.evaluate()` does not modify the examples, rerun the
pipeline on the (limited) texts in order to provide the predicted
annotation in the displacy output option.
* Add test for #7035
* Update test for issue 7056
* Fix test
* Fix transitions method used in testing
* Fix state eol detection when rebuffer
* Clean up redundant fix
* Add regression test
* Run PhraseMatcher on Spans
* Add test for PhraseMatcher on Spans and Docs
* Add SCA
* Add test with 3 matches in Doc, 1 match in Span
* Update docs
* Use doc.length for find_matches in tokenizer
Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
Now that the initialize step is fully implemented, the source of E923 is
typically missing or improperly converted/formatted data rather than a
bug in spaCy, so rephrase the error and message and remove the prompt to
open an issue.
```python
def test_vocab_lexeme_add_flag_auto_id(en_vocab):
is_len4 = en_vocab.add_flag(lambda string: len(string) == 4)
assert en_vocab["1999"].check_flag(is_len4) is True
assert en_vocab["1999"].check_flag(IS_DIGIT) is True
assert en_vocab["199"].check_flag(is_len4) is False
> assert en_vocab["199"].check_flag(IS_DIGIT) is True
E assert False is True
E + where False = <built-in method check_flag of spacy.lexeme.Lexeme object at 0x7fa155c36840>(3)
E + where <built-in method check_flag of spacy.lexeme.Lexeme object at 0x7fa155c36840> = <spacy.lexeme.Lexeme object at 0x7fa155c36840>.check_flag
spacy/tests/vocab_vectors/test_lexeme.py:49: AssertionError
```
> `pytest==6.1.1`
>
> `numpy==1.19.2`
>
> `Python version: 3.8.3`
To reproduce the error, run `pytest --random-order-bucket=global --random-order-seed=170158 -v spacy/tests`
If `test_vocab_lexeme_add_flag_auto_id` is run after `test_vocab_lexeme_add_flag_provided_id`, it fails.
It seems like `test_vocab_lexeme_add_flag_provided_id` uses the `IS_DIGIT` bit for testing purposes but does not reset the bit.
This solution seems to work but, if anyone has a better fix, please let me know and I will integrate it.
* add capture argument to project_run and run_commands
* git bump to 3.0.1
* Set version to 3.0.1.dev0
Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
When `--no-cache-dir` is present, it prevents caching to properly function.
If the user still wants to do this, there is the possibility to pass options with `user_pip_args`.
But you should not enforce options like these. In my case this is preventing some docker build (using buildkit caching) to have proper caching of models.
Instead of silently using only the first token in each matched span:
* Forbid `OP: ?/*/+` through `DependencyMatcher` validation
* As a fail-safe, add warning if a token match that's not exactly one
token long is found by a token pattern.
* add error handler for pipe methods
* add unit tests
* remove pipe method that are the same as their base class
* have Language keep track of a default error handler
* cleanup
* formatting
* small refactor
* add documentation
* Initial Spanish lemmatizer
* Handle merged verb+pron(s) multi-word tokens
* Use VERB for AUX rule lookup
* Add morph to lemma cache key
* Fix aux lookups, minor refactoring
* Improve verb+pron handling
* Move verb+pron handling into its own method
* Check for exceptions (primarily for se)
* Collect pronouns in the same (not reversed) order
* Only add modified possible lemmas
* Fix `spacy.util.minibatch` when the size iterator is finished (#6745)
* Skip 0-length matches (#6759)
Add hack to prevent matcher from returning 0-length matches.
* support IS_SENT_START in PhraseMatcher (#6771)
* support IS_SENT_START in PhraseMatcher
* add unit test and friendlier error
* use IDS.get instead
* ensure span.text works for an empty span (#6772)
* Remove unicode_literals
Co-authored-by: Santiago Castro <bryant@montevideo.com.uy>
Co-authored-by: Sofie Van Landeghem <svlandeg@users.noreply.github.com>
* Allow output_path to be None during training
* Fix cat scoring (?)
* Improve error message for weighted None score
* Improve messages
So we can call this in other places etc.
* FIx output path check
* Use latest wasabi
* Revert "Improve error message for weighted None score"
This reverts commit 70599267635e2cfcc6c8922e3e4fb20dc978beb6.
* Exclude None scores from final score by default
It's otherwise very difficult to keep track of the score weights if we modify a config programmatically, source components etc.
* Update warnings and use logger.warning
* Spacy Cli info method causing backward compatibility issues #6791
fix backward compatibility by setting default value to exclude in info
method.
* setting empty list as default argument is dangerous.
so setting default to None and then setting it to emptylist, if None.
Reference : https://nikos7am.com/posts/mutable-default-arguments/
* Adding contributor agreement for user werew
* [DependencyMatcher] Comment and clean code
* [DependencyMatcher] Use defaultdicts
* [DependencyMatcher] Simplify _retrieve_tree method
* [DependencyMatcher] Remove prepended underscores
* [DependencyMatcher] Address TODO and move grouping of token's positions out of the loop
* [DependencyMatcher] Remove _nodes attribute
* [DependencyMatcher] Use enumerate in _retrieve_tree method
* [DependencyMatcher] Clean unused vars and use camel_case naming
* [DependencyMatcher] Memoize node+operator map
* Add root property to Token
* [DependencyMatcher] Groups matches by root
* [DependencyMatcher] Remove unused _keys_to_token attribute
* [DependencyMatcher] Use a list to map tokens to matcher's keys
* [DependencyMatcher] Remove recursion
* [DependencyMatcher] Use a generator to retrieve matches
* [DependencyMatcher] Remove unused memory pool
* [DependencyMatcher] Hide private methods and attributes
* [DependencyMatcher] Improvements to the matches validation
* Apply suggestions from code review
Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
* [DependencyMatcher] Fix keys_to_position_maps
* Remove Token.root property
* [DependencyMatcher] Remove functools' lru_cache
Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com>
* warn when frozen components break listener pattern
* few notes in the documentation
* update arg name
* formatting
* cleanup
* specify listeners return type
* Avoid a SyntaxError in self-attentive-parser
Fix a usage of quotation marks in the example of spaCy Universe self-attentive-parser
* Create forest1988.md
Fill in the spaCy contributor agreement
* Adding Mindmeld to Universe JSON
Mindmeld is a conversational AI platform for deep-domain voice interfaces and chatbots. https://www.mindmeld.com/
* Signing contribution agreement.
Co-authored-by: kunshar2 <kunshar2@cisco.com>
* fix the wrong hash url in adding-languages.md file
change the #101 url hash path to #language-data
* filled in the spaCy Contributor Agreement
filled in the spaCy Contributor Agreement
about: Did you come across a bug or unexpected behaviour differing from the docs?
name: "\U0001F6A8 Submit a Bug Report"
about: Use this template if you came across a bug or unexpected behaviour differing from the docs.
---
<!-- NOTE: For questions or install related issues, please open a Discussion instead. -->
## How to reproduce the behaviour
<!-- Include a code example or the steps that led to the problem. Please try to be as specific as possible. -->
## Your Environment
<!-- Include details of your environment. If you're using spaCy 1.7+, you can also type `python -m spacy info --markdown` and copy-paste the result here.-->
<!-- Include details of your environment. You can also type `python -m spacy info --markdown` and copy-paste the result here.-->
about: Do you have problems installing spaCy, and none of the suggestions in the docs
and other issues helped?
---
<!-- Before submitting an issue, make sure to check the docs and closed issues to see if any of the solutions work for you. Installation problems can often be related to Python environment issues and problems with compilation. -->
## How to reproduce the problem
<!-- Include the details of how the problem occurred. Which command did you run to install spaCy? Did you come across an error? What else did you try? -->
```bash
# copy-paste the error message here
```
## Your Environment
<!-- Include details of your environment. If you're using spaCy 1.7+, you can also type `python -m spacy info --markdown` and copy-paste the result here.-->
about: For feature and project ideas, general usage questions or help with your code, please post on the GitHub Discussions board instead.
---
<!-- Describe your issue here. Please keep in mind that the GitHub issue tracker is mostly intended for reports related to the spaCy code base and source, and for bugs and enhancements. If you're looking for help with your code, consider posting a question here:
<!-- Include details of your environment. If you're using spaCy 1.7+, you can also type `python -m spacy info --markdown` and copy-paste the result here.-->
"message": "This issue has been automatically closed because it was answered and there was no follow-up discussion.",
"remove_label_on_comment": true,
"remove_label_on_close": true
},
"more-info-needed": {
"delay": "P7D",
"message": "This issue has been automatically closed because there has been no response to a request for more information from the original author. With only the information that is currently in the issue, there's not enough information to take action. If you're the original author, feel free to reopen the issue if you have or find the answers needed to investigate further.",
and add your own attributes, properties and methods to the `Doc`, `Token` and
`Span`. If you're looking to implement a new spaCy feature, starting with a
@@ -109,8 +106,8 @@ questions:
library later.
- **Would the feature be easier to implement if it relied on "heavy" dependencies spaCy doesn't currently require?**
Python has a very rich ecosystem. Libraries like scikit-learn, SciPy, Gensim or
TensorFlow/Keras do lots of useful things — but we don't want to have them as
Python has a very rich ecosystem. Libraries like PyTorch, TensorFlow, scikit-learn, SciPy or Gensim
do lots of useful things — but we don't want to have them as default
dependencies. If the feature requires functionality in one of these libraries,
it's probably better to break it out into a different package.
@@ -137,61 +134,61 @@ files, a compiler, [pip](https://pip.pypa.io/en/latest/installing/),
[virtualenv](https://virtualenv.pypa.io/en/stable/) and
[git](https://git-scm.com) installed. The compiler is usually the trickiest part.
```
python -m pip install -U pip
git clone https://github.com/explosion/spaCy
cd spaCy
python -m venv .env
source .env/bin/activate
export PYTHONPATH=`pwd`
pip install -r requirements.txt
python setup.py build_ext --inplace
```
If you've made changes to `.pyx` files, you need to recompile spaCy before you
If you've made changes to `.pyx` files, you need to **recompile spaCy** before you
can test your changes by re-running `python setup.py build_ext --inplace`.
Changes to `.py` files will be effective immediately.
📖 **For more details and instructions, see the documentation on [compiling spaCy from source](https://spacy.io/usage/#source) and the [quickstart widget](https://spacy.io/usage/#section-quickstart) to get the right commands for your platform and Python version.**
### Contributor agreement
If you've made a contribution to spaCy, you should fill in the
[spaCy contributor agreement](.github/CONTRIBUTOR_AGREEMENT.md) to ensure that
your contribution can be used across the project. If you agree to be bound by
the terms of the agreement, fill in the [template](.github/CONTRIBUTOR_AGREEMENT.md)
and include it with your pull request, or submit it separately to
[`.github/contributors/`](/.github/contributors). The name of the file should be
your GitHub username, with the extension `.md`. For example, the user
example_user would create the file `.github/contributors/example_user.md`.
### Fixing bugs
When fixing a bug, first create an
[issue](https://github.com/explosion/spaCy/issues) if one does not already exist.
The description text can be very short – we don't want to make this too
[issue](https://github.com/explosion/spaCy/issues) if one does not already
exist. The description text can be very short – we don't want to make this too
bureaucratic.
Next, create a test file named `test_issue[ISSUE NUMBER].py` in the
[`spacy/tests/regression`](spacy/tests/regression)folder. Test for the bug
you're fixing, and make sure the test fails. Next, add and commit your test file
referencing the issue number in the commit message. Finally, fix the bug, make
sure your test passes and reference the issue in your commit message.
Next, add a test to the relevant file in the
[`spacy/tests`](spacy/tests)folder. Then add a [pytest
spaCy is a library for advanced Natural Language Processing in Python and
spaCy is a library for **advanced Natural Language Processing** in Python and
Cython. It's built on the very latest research, and was designed from day one to
be used in real products.
spaCy comes with
[pretrained pipelines](https://spacy.io/models) and vectors, and
currently supports tokenization for **60+ languages**. It features
state-of-the-art speed, convolutional **neural network models** for tagging,
parsing, **named entity recognition**, **text classification** and more, multi-task learning with pretrained **transformers** like BERT, as well as a production-ready training system and easy model packaging, deployment and workflow management.
spaCy is commercial open-source software, released under the MIT license.
spaCy comes with [pretrained pipelines](https://spacy.io/models) and currently
supports tokenization and training for **70+ languages**. It features
state-of-the-art speed and **neural network models** for tagging, parsing,
**named entity recognition**, **text classification** and more, multi-task
learning with pretrained **transformers** like BERT, as well as a
production-ready [**training system**](https://spacy.io/usage/training) and easy
model packaging, deployment and workflow management. spaCy is commercial
| ⭐️ **[spaCy 101]** | New to spaCy? Here's everything you need to know! |
| 📚 **[Usage Guides]** | How to use spaCy and its features. |
| 🚀 **[New in v3.0]** | New features, backwards incompatibilities and migration guide. |
| 🪐 **[Project Templates]** | End-to-end workflows you can clone, modify and run. |
| 🎛 **[API Reference]** | The detailed reference for spaCy's API. |
| ⏩ **[GPU Processing]** | Use spaCy with CUDA-compatible GPU processing. |
| 📦 **[Models]** | Download trained pipelines for spaCy. |
| 🦙 **[Large Language Models]** | Integrate LLMs into spaCy pipelines. |
| 🌌 **[Universe]** | Plugins, extensions, demos and books from the spaCy ecosystem. |
| ⚙️ **[spaCy VS Code Extension]** | Additional tooling and features for working with spaCy's config files. |
| 👩🏫 **[Online Course]** | Learn spaCy in this free and interactive online course. |
| 📰 **[Blog]** | Read about current spaCy and Prodigy development, releases, talks and more from Explosion. |
| 📺 **[Videos]** | Our YouTube channel with video tutorials, talks and more. |
| 🔴 **[Live Stream]** | Join Matt as he works on spaCy and chat about NLP, live every week. |
| 🛠 **[Changelog]** | Changes and version history. |
| 💝 **[Contribute]** | How to contribute to the spaCy project and code base. |
| 👕 **[Swag]** | Support us and our work with unique, custom-designed swag! |
| <a href="https://explosion.ai/tailored-solutions"><img src="https://github.com/explosion/spaCy/assets/13643239/36d2a42e-98c0-4599-90e1-788ef75181be" width="150" alt="Tailored Solutions"/></a> | Custom NLP consulting, implementation and strategic advice by spaCy’s core development team. Streamlined, production-ready, predictable and maintainable. Send us an email or take our 5-minute questionnaire, and well'be in touch! **[Learn more →](https://explosion.ai/tailored-solutions)** |
[spacy 101]: https://spacy.io/usage/spacy-101
[new in v3.0]: https://spacy.io/usage/v3
[usage guides]: https://spacy.io/usage/
[api reference]: https://spacy.io/api/
[gpu processing]: https://spacy.io/usage#gpu
[models]: https://spacy.io/models
[large language models]: https://spacy.io/usage/large-language-models
[universe]: https://spacy.io/universe
[spacy vs code extension]: https://github.com/explosion/spacy-vscode
- **Trained pipelines** for different languages and tasks
- Multi-task learning with pretrained **transformers** like BERT
-Pretrained **word vectors**
-Support for pretrained **word vectors** and embeddings
- State-of-the-art speed
- Production-ready **training system**
- Linguistically-motivated **tokenization**
- Components for named **entity recognition**, part-of-speech-tagging, dependency parsing, sentence segmentation, **text classification**, lemmatization, morphological analysis, entity linking and more
- Components for named **entity recognition**, part-of-speech-tagging,
| **Mac** | Install a recent version of [XCode](https://developer.apple.com/xcode/), including the so-called "Command Line Tools". macOS and OS X ship with Python and git preinstalled. |
| **Windows** | Install a version of the [Visual C++ Build Tools](https://visualstudio.microsoft.com/visual-cpp-build-tools/) or [Visual Studio Express](https://visualstudio.microsoft.com/vs/express/) that matches the version that was used to compile your Python interpreter. |
For more details and instructions, see the documentation on
[compiling spaCy from source](https://spacy.io/usage#source) and the
[quickstart widget](https://spacy.io/usage#section-quickstart) to get the right
For a general overview of code conventions for contributors, see the [section in the contributing guide](https://github.com/explosion/spaCy/blob/master/CONTRIBUTING.md#code-conventions).
1. [Code compatibility](#code-compatibility)
2. [Auto-formatting](#auto-formatting)
3. [Linting](#linting)
4. [Documenting code](#documenting-code)
5. [Type hints](#type-hints)
6. [Structuring logic](#structuring-logic)
7. [Naming](#naming)
8. [Error handling](#error-handling)
9. [Writing tests](#writing-tests)
## Code compatibility
spaCy supports **Python 3.6** and above, so all code should be written compatible with 3.6. This means that there are certain new syntax features that we won't be able to use until we drop support for older Python versions. Some newer features provide backports that we can conditionally install for older versions, although we only want to do this if it's absolutely necessary. If we need to use conditional imports based on the Python version or other custom compatibility-specific helpers, those should live in `compat.py`.
## Auto-formatting
spaCy uses `black` for auto-formatting (which is also available as a pre-commit hook). It's recommended to configure your editor to perform this automatically, either triggered manually or whenever you save a file. We also have a GitHub action that regularly formats the code base and submits a PR if changes are available. Note that auto-formatting is currently only available for `.py` (Python) files, not for `.pyx` (Cython).
As a rule of thumb, if the auto-formatting produces output that looks messy, it can often indicate that there's a better way to structure the code to make it more concise.
In some specific cases, e.g. in the tests, it can make sense to disable auto-formatting for a specific block. You can do this by wrapping the code in `# fmt: off` and `# fmt: on`:
```diff
+ # fmt: off
text = "I look forward to using Thingamajig. I've been told it will make my life easier..."
[`flake8`](http://flake8.pycqa.org/en/latest/) is a tool for enforcing code style. It scans one or more files and outputs errors and warnings. This feedback can help you stick to general standards and conventions, and can be very useful for spotting potential mistakes and inconsistencies in your code. Code you write should be compatible with our flake8 rules and not cause any warnings.
```bash
flake8 spacy
```
The most common problems surfaced by linting are:
- **Trailing or missing whitespace.** This is related to formatting and should be fixed automatically by running `black`.
- **Unused imports.** Those should be removed if the imports aren't actually used. If they're required, e.g. to expose them so they can be imported from the given module, you can add a comment and `# noqa: F401` exception (see details below).
- **Unused variables.** This can often indicate bugs, e.g. a variable that's declared and not correctly passed on or returned. To prevent ambiguity here, your code shouldn't contain unused variables. If you're unpacking a list of tuples and end up with variables you don't need, you can call them `_` to indicate that they're unused.
- **Redefinition of function.** This can also indicate bugs, e.g. a copy-pasted function that you forgot to rename and that now replaces the original function.
- **Repeated dictionary keys.** This either indicates a bug or unnecessary duplication.
- **Comparison with `True`, `False`, `None`**. This is mostly a stylistic thing: when checking whether a value is `True`, `False` or `None`, you should be using `is` instead of `==`. For example, `if value is None`.
### Ignoring linter rules for special cases
To ignore a given line, you can add a comment like `# noqa: F401`, specifying the code of the error or warning we want to ignore. It's also possible to ignore several comma-separated codes at once, e.g. `# noqa: E731,E123`. In general, you should always **specify the code(s)** you want to ignore – otherwise, you may end up missing actual problems.
```python
# The imported class isn't used in this file, but imported here, so it can be
# imported *from* here by another module.
from.submoduleimportSomeClass# noqa: F401
try:
do_something()
except:# noqa: E722
# This bare except is justified, for some specific reason
do_something_else()
```
## Documenting code
All functions and methods you write should be documented with a docstring inline. The docstring can contain a simple summary, and an overview of the arguments and their (simplified) types. Modern editors will show this information to users when they call the function or method in their code.
If it's part of the public API and there's a documentation section available, we usually add the link as `DOCS:` at the end. This allows us to keep the docstrings simple and concise, while also providing additional information and examples if necessary.
```python
defhas_pipe(self,name:str)->bool:
"""Check if a component name is present in the pipeline. Equivalent to
`name in nlp.pipe_names`.
name (str): Name of the component.
RETURNS (bool): Whether a component of the name exists in the pipeline.
DOCS: https://spacy.io/api/language#has_pipe
"""
...
```
We specifically chose this approach of maintaining the docstrings and API reference separately, instead of auto-generating the API docs from the docstrings like other packages do. We want to be able to provide extensive explanations and examples in the documentation and use our own custom markup for it that would otherwise clog up the docstrings. We also want to be able to update the documentation independently of the code base. It's slightly more work, but it's absolutely worth it in terms of user and developer experience.
### Inline code comments
We don't expect you to add inline comments for everything you're doing – this should be obvious from reading the code. If it's not, the first thing to check is whether your code can be improved to make it more explicit. That said, if your code includes complex logic or aspects that may be unintuitive at first glance (or even included a subtle bug that you ended up fixing), you should leave a quick comment that provides more context.
```diff
token_index = indices[value]
+ # Index describes Token.i of last token but Span indices are inclusive
span = doc[prev_token_index:token_index + 1]
```
```diff
+ # To create the components we need to use the final interpolated config
+ # so all values are available (if component configs use variables).
+ # Later we replace the component config with the raw config again.
interpolated = filled.interpolate() if not filled.is_interpolated else filled
```
Don't be shy about including comments for tricky parts that _you_ found hard to implement or get right – those may come in handy for the next person working on this code, or even future you!
If your change implements a fix to a specific issue, it can often be helpful to include the issue number in the comment, especially if it's a relatively straightforward adjustment:
```diff
+ # Ensure object is a Span, not a Doc (#1234)
if isinstance(obj, Doc):
obj = obj[obj.start:obj.end]
```
### Including TODOs
It's fine to include code comments that indicate future TODOs, using the `TODO:` prefix. Modern editors typically format this in a different color, so it's easy to spot. TODOs don't necessarily have to be things that are absolutely critical to fix fight now – those should already be addressed in your pull request once it's ready for review. But they can include notes about potential future improvements.
```diff
+ # TODO: this is currently pretty slow
dir_checksum = hashlib.md5()
for sub_file in sorted(fp for fp in path.rglob("*") if fp.is_file()):
dir_checksum.update(sub_file.read_bytes())
```
If any of the TODOs you've added are important and should be fixed soon, you should add a task for this on Explosion's internal Ora board or an issue on the public issue tracker to make sure we don't forget to address it.
## Type hints
We use Python type hints across the `.py` files wherever possible. This makes it easy to understand what a function expects and returns, and modern editors will be able to show this information to you when you call an annotated function. Type hints are not currently used in the `.pyx` (Cython) code, except for definitions of registered functions and component factories, where they're used for config validation. Ideally when developing, run `mypy spacy` on the code base to inspect any issues.
If possible, you should always use the more descriptive type hints like `List[str]` or even `List[Any]` instead of only `list`. We also annotate arguments and return types of `Callable`– although, you can simplify this if the type otherwise gets too verbose (e.g. functions that return factories to create callbacks). Remember that `Callable` takes two values: a **list** of the argument type(s) in order, and the return values.
For typing variables, we prefer the explicit format.
```diff
- var = value # type: Type
+ var: Type = value
```
For model architectures, Thinc also provides a collection of [custom types](https://thinc.ai/docs/api-types), including more specific types for arrays and model inputs/outputs. Even outside of static type checking, using these types will make the code a lot easier to read and follow, since it's always clear what array types are expected (and what might go wrong if the output is different from the expected type).
If you need to use a type hint that refers to something later declared in the same module, or the class that a method belongs to, you can use a string value instead:
```python
classSomeClass:
deffrom_bytes(self,data:bytes)->"SomeClass":
...
```
In some cases, you won't be able to import a class from a different module to use it as a type hint because it'd cause circular imports. For instance, `spacy/util.py` includes various helper functions that return an instance of `Language`, but we couldn't import it, because `spacy/language.py` imports `util` itself. In this case, we can provide `"Language"` as a string and make the import conditional on `typing.TYPE_CHECKING` so it only runs when the code is evaluated by a type checker:
```python
fromtypingTYPE_CHECKING
ifTYPE_CHECKING:
from.languageimportLanguage
defload_model(name:str)->"Language":
...
```
Note that we typically put the `from typing` import statements on the first line(s) of the Python module.
## Structuring logic
### Positional and keyword arguments
We generally try to avoid writing functions and methods with too many arguments, and use keyword-only arguments wherever possible. Python lets you define arguments as keyword-only by separating them with a `, *`. If you're writing functions with additional arguments that customize the behavior, you typically want to make those arguments keyword-only, so their names have to be provided explicitly.
This makes the function calls easier to read, because it's immediately clear what the additional values mean. It also makes it easier to extend arguments or change their order later on, because you don't end up with any function calls that depend on a specific positional order.
### Avoid mutable default arguments
A common Python gotcha are [mutable default arguments](https://docs.python-guide.org/writing/gotchas/#mutable-default-arguments): if your argument defines a mutable default value like `[]` or `{}` and then goes and mutates it, the default value is created _once_ when the function is created and the same object is then mutated every time the function is called. This can be pretty unintuitive when you first encounter it. We therefore avoid writing logic that does this.
If your arguments need to default to an empty list or dict, you can use the `SimpleFrozenList` and `SimpleFrozenDict` helpers provided by spaCy. They are simple frozen implementations that raise an error if they're being mutated to prevent bugs and logic that accidentally mutates default arguments.
We strongly discourage using `try`/`except` blocks for anything that's not third-party error handling or error handling that we otherwise have little control over. There's typically always a way to anticipate the _actual_ problem and **check for it explicitly**, which makes the code easier to follow and understand, and prevents bugs:
```diff
- try:
- token = doc[i]
- except IndexError:
- token = doc[-1]
+ if i < len(doc):
+ token = doc[i]
+ else:
+ token = doc[-1]
```
Even if you end up having to check for multiple conditions explicitly, this is still preferred over a catch-all `try`/`except`. It can be very helpful to think about the exact scenarios you need to cover, and what could go wrong at each step, which often leads to better code and fewer bugs. `try/except` blocks can also easily mask _other_ bugs and problems that raise the same errors you're catching, which is obviously bad.
If you have to use `try`/`except`, make sure to only include what's **absolutely necessary** in the `try` block and define the exception(s) explicitly. Otherwise, you may end up masking very different exceptions caused by other bugs.
`lambda` functions can be useful for defining simple anonymous functions in a single line, but they also introduce problems: for instance, they require [additional logic](https://stackoverflow.com/questions/25348532/can-python-pickle-lambda-functions) in order to be pickled and are pretty ugly to type-annotate. So we typically avoid them in the code base and only use them in the serialization handlers and within tests for simplicity. Instead of `lambda`s, check if your code can be refactored to not need them, or use helper functions instead.
```diff
- split_string: Callable[[str], List[str]] = lambda value: [v.strip() for v in value.split(",")]
+ def split_string(value: str) -> List[str]:
+ return [v.strip() for v in value.split(",")]
```
### Numeric comparisons
For numeric comparisons, as a general rule we always use `<` and `>=` and avoid the usage of `<=` and `>`. This is to ensure we consistently
apply inclusive lower bounds and exclusive upper bounds, helping to prevent off-by-one errors.
One exception to this rule is the ternary case. With a chain like
```python
ifvalue>=0andvalue<max:
...
```
it's fine to rewrite this to the shorter form
```python
if0<=value<max:
...
```
even though this requires the usage of the `<=` operator.
### Iteration and comprehensions
We generally avoid using built-in functions like `filter` or `map` in favor of list or generator comprehensions.
```diff
- filtered = filter(lambda x: x in ["foo", "bar"], values)
+ filtered = (x for x in values if x in ["foo", "bar"])
- filtered = list(filter(lambda x: x in ["foo", "bar"], values))
+ filtered = [x for x in values if x in ["foo", "bar"]]
- result = map(lambda x: { x: x in ["foo", "bar"]}, values)
+ result = ({x: x in ["foo", "bar"]} for x in values)
- result = list(map(lambda x: { x: x in ["foo", "bar"]}, values))
+ result = [{x: x in ["foo", "bar"]} for x in values]
```
If your logic is more complex, it's often better to write a loop instead, even if it adds more lines of code in total. The result will be much easier to follow and understand.
```diff
- result = [{"key": key, "scores": {f"{i}": score for i, score in enumerate(scores)}} for key, scores in values]
+ result = []
+ for key, scores in values:
+ scores_dict = {f"{i}": score for i, score in enumerate(scores)}
Although spaCy uses a lot of classes, **inheritance is viewed with some suspicion** — it's seen as a mechanism of last resort. You should discuss plans to extend the class hierarchy before implementing. Unless you're implementing a new data structure or pipeline component, you typically shouldn't have to use classes at all.
### Don't use `print`
The core library never `print`s anything. While we encourage using `print` statements for simple debugging (it's the most straightforward way of looking at what's happening), make sure to clean them up once you're ready to submit your pull request. If you want to output warnings or debugging information for users, use the respective dedicated mechanisms for this instead (see sections on warnings and logging for details).
The only exceptions are the CLI functions, which pretty-print messages for the user, and methods that are explicitly intended for printing things, e.g. `Language.analyze_pipes` with `pretty=True` enabled. For this, we use our lightweight helper library [`wasabi`](https://github.com/ines/wasabi).
## Naming
Naming is hard and often a topic of long internal discussions. We don't expect you to come up with the perfect names for everything you write – finding the right names is often an iterative and collaborative process. That said, we do try to follow some basic conventions.
Consistent with general Python conventions, we use `CamelCase` for class names including dataclasses, `snake_case` for methods, functions and variables, and `UPPER_SNAKE_CASE` for constants, typically defined at the top of a module. We also avoid using variable names that shadow the names of built-in functions, e.g. `input`, `help` or `list`.
### Naming variables
Variable names should always make it clear _what exactly_ the variable is and what it's used for. Instances of common classes should use the same consistent names. For example, you should avoid naming a text string (or anything else that's not a `Doc` object) `doc`. The most common class-to-variable mappings are:
We try to avoid introducing too many temporary variables, as these clutter your namespace. It's okay to re-assign to an existing variable, but only if the value has the same type.
```diff
ents = get_a_list_of_entities()
ents = [ent for ent in doc.ents if ent.label_ == "PERSON"]
- ents = {(ent.start, ent.end): ent.label_ for ent in ents}
+ ent_mappings = {(ent.start, ent.end): ent.label_ for ent in ents}
```
### Naming methods and functions
Try choosing short and descriptive names wherever possible and imperative verbs for methods that do something, e.g. `disable_pipes`, `add_patterns` or `get_vector`. Private methods and functions that are not intended to be part of the user-facing API should be prefixed with an underscore `_`. It's often helpful to look at the existing classes for inspiration.
Objects that can be serialized, e.g. data structures and pipeline components, should implement the same consistent methods for serialization. Those usually include at least `to_disk`, `from_disk`, `to_bytes` and `from_bytes`. Some objects can also implement more specific methods like `{to/from}_dict` or `{to/from}_str`.
## Error handling
We always encourage writing helpful and detailed custom error messages for everything we can anticipate going wrong, and including as much detail as possible. spaCy provides a directory of error messages in `errors.py` with unique codes for each message. This allows us to keep the code base more concise and avoids long and nested blocks of texts throughout the code that disrupt the reading flow. The codes make it easy to find references to the same error in different places, and also helps identify problems reported by users (since we can just search for the error code).
Errors can be referenced via their code, e.g. `Errors.E123`. Messages can also include placeholders for values, that can be populated by formatting the string with `.format()`.
As a general rule of thumb, all error messages raised within the **core library** should be added to `Errors`. The only place where we write errors and messages as strings is `spacy.cli`, since these functions typically pretty-print and generate a lot of output that'd otherwise be very difficult to separate from the actual logic.
### Re-raising exceptions
If we anticipate possible errors in third-party code that we don't control, or our own code in a very different context, we typically try to provide custom and more specific error messages if possible. If we need to re-raise an exception within a `try`/`except` block, we can re-raise a custom exception.
[Re-raising `from`](https://docs.python.org/3/tutorial/errors.html#exception-chaining) the original caught exception lets us chain the exceptions, so the user sees both the original error, as well as the custom message with a note "The above exception was the direct cause of the following exception".
```diff
try:
run_third_party_code_that_might_fail()
except ValueError as e:
+ raise ValueError(Errors.E123) from e
```
In some cases, it makes sense to suppress the original exception, e.g. if we know what it is and know that it's not particularly helpful. In that case, we can raise `from None`. This prevents clogging up the user's terminal with multiple and irrelevant chained exceptions.
```diff
try:
run_our_own_code_that_might_fail_confusingly()
except ValueError:
+ raise ValueError(Errors.E123) from None
```
### Avoid using naked `assert`
During development, it can sometimes be helpful to add `assert` statements throughout your code to make sure that the values you're working with are what you expect. However, as you clean up your code, those should either be removed or replaced by more explicit error handling:
Otherwise, the user will get to see a naked `AssertionError` with no further explanation, which is very unhelpful. Instead of adding an error message to `assert`, it's always better to `raise` more explicit errors for specific conditions. If you're checking for something that _has to be right_ and would otherwise be a bug in spaCy, you can express this in the error message:
```python
E161=("Found an internal inconsistency when predicting entity links. "
"This is likely a bug in spaCy, so feel free to open an issue: "
"https://github.com/explosion/spaCy/issues")
```
### Warnings
Instead of raising an error, some parts of the code base can raise warnings to notify the user of a potential problem. This is done using Python's `warnings.warn` and the messages defined in `Warnings` in the `errors.py`. Whether or not warnings are shown can be controlled by the user, including custom filters for disabling specific warnings using a regular expression matching our internal codes, e.g. `W123`.
```diff
- print("Warning: No examples provided for validation")
+ warnings.warn(Warnings.W123)
```
When adding warnings, make sure you're not calling `warnings.warn` repeatedly, e.g. in a loop, which will clog up the terminal output. Instead, you can collect the potential problems first and then raise a single warning. If the problem is critical, consider raising an error instead.
Log statements can be added via spaCy's `logger`, which uses Python's native `logging` module under the hood. We generally only use logging for debugging information that **the user may choose to see** in debugging mode or that's **relevant during training** but not at runtime.
```diff
+ logger.info("Set up nlp object from config")
config = nlp.config.interpolate()
```
`spacy train` and similar CLI commands will enable all log statements of level `INFO` by default (which is not the case at runtime). This allows outputting specific information within certain parts of the core library during training, without having it shown at runtime. `DEBUG`-level logs are only shown if the user enables `--verbose` logging during training. They can be used to provide more specific and potentially more verbose details, especially in areas that can indicate bugs or problems, or to surface more details about what spaCy does under the hood. You should only use logging statements if absolutely necessary and important.
## Writing tests
spaCy uses the [`pytest`](http://doc.pytest.org/) framework for testing. Tests for spaCy modules and classes live in their own directories of the same name and all test files should be prefixed with `test_`. Tests included in the core library only cover the code and do not depend on any trained pipelines. When implementing a new feature or fixing a bug, it's usually good to start by writing some tests that describe what _should_ happen. As you write your code, you can then keep running the relevant tests until all of them pass.
### Test suite structure
When adding tests, make sure to use descriptive names and only test for one behavior at a time. Tests should be grouped into modules dedicated to the same type of functionality and some test modules are organized as directories of test files related to the same larger area of the library, e.g. `matcher` or `tokenizer`.
Regression tests are tests that refer to bugs reported in specific issues. They should live in the relevant module of the test suite, named according to the issue number (e.g., `test_issue1234.py`), and [marked](https://docs.pytest.org/en/6.2.x/example/markers.html#working-with-custom-markers) appropriately (e.g. `@pytest.mark.issue(1234)`). This system allows us to relate tests for specific bugs back to the original reported issue, which is especially useful if we introduce a regression and a previously passing regression tests suddenly fails again. When fixing a bug, it's often useful to create a regression test for it first.
The test suite also provides [fixtures](https://github.com/explosion/spaCy/blob/master/spacy/tests/conftest.py) for different language tokenizers that can be used as function arguments of the same name and will be passed in automatically. Those should only be used for tests related to those specific languages. We also have [test utility functions](https://github.com/explosion/spaCy/blob/master/spacy/tests/util.py) for common operations, like creating a temporary file.
### Testing Cython Code
If you're developing Cython code (`.pyx` files), those extensions will need to be built before the test runner can test that code - otherwise it's going to run the tests with stale code from the last time the extension was built. You can build the extensions locally with `python setup.py build_ext -i`.
### Constructing objects and state
Test functions usually follow the same simple structure: they set up some state, perform the operation you want to test and `assert` conditions that you expect to be true, usually before and after the operation.
Tests should focus on exactly what they're testing and avoid dependencies on other unrelated library functionality wherever possible. If all your test needs is a `Doc` object with certain annotations set, you should always construct it manually:
If you need to run the same test function over different input examples, you usually want to parametrize the test cases instead of using a loop within your test. This lets you keep a better separation between test cases and test logic, and it'll result in more useful output because `pytest` will be able to tell you which exact test case failed.
The `@pytest.mark.parametrize` decorator takes two arguments: a string defining one or more comma-separated arguments that should be passed to the test function and a list of corresponding test cases (or a list of tuples to provide multiple arguments).
deftest_token_length(en_tokenizer,text,expected_len):# en_tokenizer is a fixture
doc=en_tokenizer(text)
assertlen(doc)==expected_len
```
You can also stack `@pytest.mark.parametrize` decorators, although this is not recommended unless it's absolutely needed or required for the test. When stacking decorators, keep in mind that this will run the test with all possible combinations of the respective parametrized values, which is often not what you want and can slow down the test suite.
### Handling failing tests
`xfail` means that a test **should pass but currently fails**, i.e. is expected to fail. You can mark a test as currently xfailing by adding the `@pytest.mark.xfail` decorator. This should only be used for tests that don't yet work, not for logic that cause errors we raise on purpose (see the section on testing errors for this). It's often very helpful to implement tests for edge cases that we don't yet cover and mark them as `xfail`. You can also provide a `reason` keyword argument to the decorator with an explanation of why the test currently fails.
```diff
+ @pytest.mark.xfail(reason="Issue #225 - not yet implemented")
doc = en_tokenizer("Will this road take me to Puddleton?\u2014No.")
assert doc[8].text == "\u2014"
```
When you run the test suite, you may come across tests that are reported as `xpass`. This means that they're marked as `xfail` but didn't actually fail. This is worth looking into: sometimes, it can mean that we have since fixed a bug that caused the test to previously fail, so we can remove the decorator. In other cases, especially when it comes to machine learning model implementations, it can also indicate that the **test is flaky**: it sometimes passes and sometimes fails. This can be caused by a bug, or by constraints being too narrowly defined. If a test shows different behavior depending on whether its run in isolation or not, this can indicate that it reacts to global state set in a previous test, which is unideal and should be avoided.
### Writing slow tests
If a test is useful but potentially quite slow, you can mark it with the `@pytest.mark.slow` decorator. This is a special marker we introduced and tests decorated with it only run if you run the test suite with `--slow`, but not as part of the main CI process. Before introducing a slow test, double-check that there isn't another and more efficient way to test for the behavior. You should also consider adding a simpler test with maybe only a subset of the test cases that can always run, so we at least have some coverage.
### Skipping tests
The `@pytest.mark.skip` decorator lets you skip tests entirely. You only want to do this for failing tests that may be slow to run or cause memory errors or segfaults, which would otherwise terminate the entire process and wouldn't be caught by `xfail`. We also sometimes use the `skip` decorator for old and outdated regression tests that we want to keep around but that don't apply anymore. When using the `skip` decorator, make sure to provide the `reason` keyword argument with a quick explanation of why you chose to skip this test.
### Testing errors and warnings
`pytest` lets you check whether a given error is raised by using the `pytest.raises` contextmanager. This is very useful when implementing custom error handling, so make sure you're not only testing for the correct behavior but also for errors resulting from incorrect inputs. If you're testing errors, you should always check for `pytest.raises` explicitly and not use `xfail`.
You can also use the `pytest.warns` contextmanager to check that a given warning type is raised. The first argument is the warning type or `None` (which will capture a list of warnings that you can `assert` is empty).
```python
deftest_phrase_matcher_validation(en_vocab):
doc1=Doc(en_vocab,words=["Test"],deps=["ROOT"])
doc2=Doc(en_vocab,words=["Test"])
matcher=PhraseMatcher(en_vocab,validate=True)
withpytest.warns(UserWarning):
# Warn about unnecessarily parsed document
matcher.add("TEST1",[doc1])
withpytest.warns(None)asrecord:
matcher.add("TEST2",[docs])
assertnotrecord.list
```
Keep in mind that your tests will fail if you're using the `pytest.warns` contextmanager with a given warning and the warning is _not_ shown. So you should only use it to check that spaCy handles and outputs warnings correctly. If your test outputs a warning that's expected but not relevant to what you're testing, you can use the `@pytest.mark.filterwarnings` decorator and ignore specific warnings starting with a given code:
```python
@pytest.mark.filterwarnings("ignore:\\[W036")
deftest_matcher_empty(en_vocab):
matcher=Matcher(en_vocab)
matcher(Doc(en_vocab,words=["test"]))
```
### Testing trained pipelines
Our regular test suite does not depend on any of the trained pipelines, since their outputs can vary and aren't generally required to test the library functionality. We test pipelines separately using the tests included in the [`spacy-models`](https://github.com/explosion/spacy-models) repository, which run whenever we train a new suite of models. The tests here mostly focus on making sure that the packages can be loaded and that the predictions seam reasonable, and they include checks for common bugs we encountered previously. If your test does not primarily focus on verifying a model's predictions, it should be part of the core library tests and construct the required objects manually, instead of being added to the models tests.
Keep in mind that specific predictions may change, and we can't test for all incorrect predictions reported by users. Different models make different mistakes, so even a model that's significantly more accurate overall may end up making wrong predictions that it previously didn't. However, some surprising incorrect predictions may indicate deeper bugs that we definitely want to investigate.
Explosion-bot is a robot that can be invoked to help with running particular test commands.
## Permissions
Only maintainers have permissions to summon explosion-bot. Each of the open source repos that use explosion-bot has its own team(s) of maintainers, and only github users who are members of those teams can successfully run bot commands.
## Running robot commands
To summon the robot, write a github comment on the issue/PR you wish to test. The comment must be in the following format:
```
@explosion-bot please test_gpu
```
Some things to note:
- The `@explosion-bot please` must be the beginning of the command - you cannot add anything in front of this or else the robot won't know how to parse it. Adding anything at the end aside from the test name will also confuse the robot, so keep it simple!
- The command name (such as `test_gpu`) must be one of the tests that the bot knows how to run. The available commands are documented in the bot's [workflow config](https://github.com/explosion/spaCy/blob/master/.github/workflows/explosionbot.yml#L26) and must match exactly one of the commands listed there.
- The robot can't do multiple things at once, so if you want it to run multiple tests, you'll have to summon it with one comment per test.
### Examples
- Execute spaCy slow GPU tests with a custom thinc branch from a spaCy PR:
This will launch the GPU pipeline for the `spacy-transformers` repo on its `master` branch, using the current spaCy PR's branch to build spaCy. The name of the repository passed to `--run-on` is case-sensitive, e.g: use `spaCy` instead of `spacy`.
- General info about supported commands.
```
@explosion-bot please info
```
- Help text for a specific command
```
@explosion-bot please <command> --help
```
## Troubleshooting
If the robot isn't responding to commands as expected, you can check its logs in the [Github Action](https://github.com/explosion/spaCy/actions/workflows/explosionbot.yml).
For each command sent to the bot, there should be a run of the `explosion-bot` workflow. In the `Install and run explosion-bot` step, towards the ends of the logs you should see info about the configuration that the bot was run with, as well as any errors that the bot encountered.
- [A. Initialization for training](#2a-initialization-for-training): `init_nlp`
- [B. Initializing the `nlp` object](#2b-initializing-the-nlp-object): `Language.initialize`
- [C. Initializing the vocab](#2c-initializing-the-vocab): `init_vocab`
## 1. Constructing the `nlp` object from a config
### 1A. Overview
Most of the functions referenced in the config are regular functions with arbitrary arguments registered via the function registry. However, the pipeline components are a bit special: they don't only receive arguments passed in via the config file, but also the current `nlp` object and the string `name` of the individual component instance (so a user can have multiple components created with the same factory, e.g. `ner_one` and `ner_two`). This name can then be used by the components to add to the losses and scores. This special requirement means that pipeline components can't just be resolved via the config the "normal" way: we need to retrieve the component functions manually and pass them their arguments, plus the `nlp` and `name`.
The `Language.from_config` classmethod takes care of constructing the `nlp` object from a config. It's the single place where this happens and what `spacy.load` delegates to under the hood. Its main responsibilities are:
- **Load and validate the config**, and optionally **auto-fill** all missing values that we either have defaults for in the config template or that registered function arguments define defaults for. This helps ensure backwards-compatibility, because we're able to add a new argument `foo: str = "bar"` to an existing function, without breaking configs that don't specity it.
- **Execute relevant callbacks** for pipeline creation, e.g. optional functions called before and after creation of the `nlp` object and pipeline.
- **Initialize language subclass and create tokenizer**. The `from_config` classmethod will always be called on a language subclass, e.g. `English`, not on `Language` directly. Initializing the subclass takes a callback to create the tokenizer.
- **Set up the pipeline components**. Components can either refer to a component factory or a `source`, i.e. an existing pipeline that's loaded and that the component is then copied from. We also need to ensure that we update the information about which components are disabled.
- **Manage listeners.** If sourced components "listen" to other components (`tok2vec`, `transformer`), we need to ensure that the references are valid. If the config specifies that listeners should be replaced by copies (e.g. to give the `ner` component its own `tok2vec` model instead of listening to the shared `tok2vec` component in the pipeline), we also need to take care of that.
Note that we only resolve and load **selected sections** in `Language.from_config`, i.e. only the parts that are relevant at runtime, which is `[nlp]` and `[components]`. We don't want to be resolving anything related to training or initialization, since this would mean loading and constructing unnecessary functions, including functions that require information that isn't necessarily available at runtime, like `paths.train`.
### 1B. How pipeline component factories work in the config
As opposed to regular registered functions that refer to a registry and function name (e.g. `"@misc": "foo.v1"`), pipeline components follow a different format and refer to their component `factory` name. This corresponds to the name defined via the `@Language.component` or `@Language.factory` decorator. We need this decorator to define additional meta information for the components, like their default config and score weights.
```ini
[components.my_component]
factory="foo"
some_arg="bar"
other_arg=${paths.some_path}
```
This means that we need to create and resolve the `config["components"]` separately from the rest of the config. There are some important considerations and things we need to manage explicitly to avoid unexpected behavior:
#### Variable interpolation
When a config is resolved, references to variables are replaced, so that the functions receive the correct value instead of just the variable name. To interpolate a config, we need it in its entirety: we couldn't just interpolate a subsection that refers to variables defined in a different subsection. So we first interpolate the entire config.
However, the `nlp.config` should include the original config with variables intact – otherwise, loading a pipeline and saving it to disk will destroy all logic implemented via variables and hard-code the values all over the place. This means that when we create the components, we need to keep two versions of the config: the interpolated config with the "real" values and the `raw_config` including the variable references.
#### Factory registry
Component factories are special and use the `@Language.factory` or `@Language.component` decorator to register themselves and their meta. When the decorator runs, it performs some basic validation, stores the meta information for the factory on the `Language` class (default config, scores etc.) and then adds the factory function to `registry.factories`. The `component` decorator can be used for registering simple functions that just take a `Doc` object and return it so in that case, we create the factory for the user automatically.
There's one important detail to note about how factories are registered via entry points: A package that wants to expose spaCy components still needs to register them via the `@Language` decorators so we have the component meta information and can perform required checks. All we care about here is that the decorated function is **loaded and imported**. When it is, the `@Language` decorator takes care of everything, including actually registering the component factory.
Normally, adding to the registry via an entry point will just add the function to the registry under the given name. But for `spacy_factories`, we don't actually want that: all we care about is that the function decorated with `@Language` is imported so the decorator runs. So we only exploit Python's entry point system to automatically import the function, and the `spacy_factories` entry point group actually adds to a **separate registry**, `registry._factories`, under the hood. Its only purpose is that the functions are imported. The decorator then runs, creates the factory if needed and adds it to the `registry.factories` registry.
#### Language-specific factories
spaCy supports registering factories on the `Language` base class, as well as language-specific subclasses like `English` or `German`. This allows providing different factories depending on the language, e.g. a different default lemmatizer. The `Language.get_factory_name` classmethod constructs the factory name as `{lang}.{name}` if a language is available (i.e. if it's a subclass) and falls back to `{name}` otherwise. So `@German.factory("foo")` will add a factory `de.foo` under the hood. If you add `nlp.add_pipe("foo")`, we first check if there's a factory for `{nlp.lang}.foo` and if not, we fall back to checking for a factory `foo`.
#### Creating a pipeline component from a factory
`Language.add_pipe` takes care of adding a pipeline component, given its factory name, its config. If no source pipeline to copy the component from is provided, it delegates to `Language.create_pipe`, which sets up the actual component function.
- Validate the config and make sure that the factory was registered via the decorator and that we have meta for it.
- Update the component config with any defaults specified by the component's `default_config`, if available. This is done by merging the values we receive into the defaults. It ensures that you can still add a component without having to specify its _entire_ config including more complex settings like `model`. If no `model` is defined, we use the default.
- Check if we have a language-specific factory for the given `nlp.lang` and if not, fall back to the global factory.
- Construct the component config, consisting of whatever arguments were provided, plus the current `nlp` object and `name`, which are default expected arguments of all factories. We also add a reference to the `@factories` registry, so we can resolve the config via the registry, like any other config. With the added `nlp` and `name`, it should now include all expected arguments of the given function.
- Fill the config to make sure all unspecified defaults from the function arguments are added and update the `raw_config` (uninterpolated with variables intact) with that information, so the component config we store in `nlp.config` is up to date. We do this by adding the `raw_config`_into_ the filled config – otherwise, the references to variables would be overwritten.
- Resolve the config and create all functions it refers to (e.g. `model`). This gives us the actual component function that we can insert into the pipeline.
### 1C. Sourcing a pipeline component
```ini
[components.ner]
source="en_core_web_sm"
```
spaCy also allows ["sourcing" a component](https://spacy.io/usage/processing-pipelines#sourced-components), which will copy it over from an existing pipeline. In this case, `Language.add_pipe` will delegate to `Language.create_pipe_from_source`. In order to copy a component effectively and validate it, the source pipeline first needs to be loaded. This is done in `Language.from_config`, so a source pipeline only has to be loaded once if multiple components source from it. Sourcing a component will perform the following checks and modifications:
- For each sourced pipeline component loaded in `Language.from_config`, a hash of the vectors data from the source pipeline is stored in the pipeline meta so we're able to check whether the vectors match and warn if not (since different vectors that are used as features in components can lead to degraded performance). Because the vectors are not loaded at the point when components are sourced, the check is postponed to `init_vocab` as part of `Language.initialize`.
- If the sourced pipeline component is loaded through `Language.add_pipe(source=)`, the vectors are already loaded and can be compared directly. The check compares the shape and keys first and finally falls back to comparing the actual byte representation of the vectors (which is slower).
- Ensure that the component is available in the pipeline.
- Interpolate the entire config of the source pipeline so all variables are replaced and the component's config that's copied over doesn't include references to variables that are not available in the destination config.
- Add the source `vocab.strings` to the destination's `vocab.strings` so we don't end up with unavailable strings in the final pipeline (which would also include labels used by the sourced component).
Note that there may be other incompatibilities that we're currently not checking for and that could cause a sourced component to not work in the destination pipeline. We're interested in adding more checks here but there'll always be a small number of edge cases we'll never be able to catch, including a sourced component depending on other pipeline state that's not available in the destination pipeline.
### 1D. Tracking components as they're modified
The `Language` class implements methods for removing, replacing or renaming pipeline components. Whenever we make these changes, we need to update the information stored on the `Language` object to ensure that it matches the current state of the pipeline. If a user just writes to `nlp.config` manually, we obviously can't ensure that the config matches the reality – but since we offer modification via the pipe methods, it's expected that spaCy keeps the config in sync under the hood. Otherwise, saving a modified pipeline to disk and loading it back wouldn't work. The internal attributes we need to keep in sync here are:
| `Language._components` | `List[Tuple[str, Callable]]` | All pipeline components as `(name, func)` tuples. This is used as the source of truth for `Language.pipeline`, `Language.pipe_names` and `Language.components`. |
| `Language._pipe_meta` | `Dict[str, FactoryMeta]` | The meta information of a component's factory, keyed by component name. This can include multiple components referring to the same factory meta. |
| `Language._pipe_configs` | `Dict[str, Config]` | The component's config, keyed by component name. |
| `Language._disabled` | `Set[str]` | Names of components that are currently disabled. |
| `Language._config` | `Config` | The underlying config. This is only internals and will be used as the basis for constructing the config in the `Language.config` property. |
In addition to the actual component settings in `[components]`, the config also allows specifying component-specific arguments via the `[initialize.components]` block, which are passed to the component's `initialize` method during initialization if it's available. So we also need to keep this in sync in the underlying config.
### 1E. spaCy's config utility functions
When working with configs in spaCy, make sure to use the utility functions provided by spaCy if available, instead of calling the respective `Config` methods. The utilities take care of providing spaCy-specific error messages and ensure a consistent order of config sections by setting the `section_order` argument. This ensures that exported configs always have the same consistent format.
-`util.load_config`: load a config from a file
-`util.load_config_from_str`: load a confirm from a string representation
-`util.copy_config`: deepcopy a config
## 2. Initialization
Initialization is a separate step of the [config lifecycle](https://spacy.io/usage/training#config-lifecycle) that's not performed at runtime. It's implemented via the `training.initialize.init_nlp` helper and calls into `Language.initialize` method, which sets up the pipeline and component models before training. The `initialize` method takes a callback that returns a sample of examples, which is used to initialize the component models, add all required labels and perform shape inference if applicable.
Components can also define custom initialization setting via the `[initialize.components]` block, e.g. if they require external data like lookup tables to be loaded in. All config settings defined here will be passed to the component's `initialize` method, if it implements one. Components are expected to handle their own serialization after they're initialized so that any data or settings they require are saved with the pipeline and will be available from disk when the pipeline is loaded back at runtime.
### 2A. Initialization for training
The `init_nlp` function is called before training and returns an initialized `nlp` object that can be updated with the examples. It only needs the config and does the following:
- Load and validate the config. In order to validate certain settings like the `seed`, we also interpolate the config to get the final value (because in theory, a user could provide this via a variable).
- Set up the GPU allocation, if required.
- Create the `nlp` object from the raw, uninterpolated config, which delegates to `Language.from_config`. Since this method may modify and auto-fill the config and pipeline component settings, we then use the interpolated version of `nlp.config` going forward, to ensure that what we're training with is up to date.
- Resolve the `[training]` block of the config and perform validation, e.g. to check that the corpora are available.
- Determine the components that should be frozen (not updated during training) or resumed (sourced components from a different pipeline that should be updated from the examples and not reset and re-initialized). To resume training, we can call the `nlp.resume_training` method.
- Initialize the `nlp` object via `nlp.initialize` and pass it a `get_examples` callback that returns the training corpus (used for shape inference, setting up labels etc.). If the training corpus is streamed, we only provide a small sample of the data, which can potentially be infinite. `nlp.initialize` will delegate to the components as well and pass the data sample forward.
- Check the listeners and warn about components dependencies, e.g. if a frozen component listens to a component that is retrained, or vice versa (which can degrade results).
### 2B. Initializing the `nlp` object
The `Language.initialize` method does the following:
- **Resolve the config** defined in the `[initialize]` block separately (since everything else is already available in the loaded `nlp` object), based on the fully interpolated config.
- **Execute callbacks**, i.e. `before_init` and `after_init`, if they're defined.
- **Initialize the vocab**, including vocab data, lookup tables and vectors.
- **Initialize the tokenizer** if it implements an `initialize` method. This is not the case for the default tokenizers, but it allows custom tokenizers to depend on external data resources that are loaded in on initialization.
- **Initialize all pipeline components** if they implement an `initialize` method and pass them the `get_examples` callback, the current `nlp` object as well as well additional initialization config settings provided in the component-specific block.
- **Initialize pretraining** if a `[pretraining]` block is available in the config. This allows loading pretrained tok2vec weights in `spacy pretrain`.
- **Register listeners** if token-to-vector embedding layers of a component model "listen" to a previous component (`tok2vec`, `transformer`) in the pipeline.
- **Create an optimizer** on the `Language` class, either by adding the optimizer passed as `sgd` to `initialize`, or by creating the optimizer defined in the config's training settings.
### 2C. Initializing the vocab
Vocab initialization is handled in the `training.initialize.init_vocab` helper. It takes the relevant loaded functions and values from the config and takes care of the following:
- Add lookup tables defined in the config initialization, e.g. custom lemmatization tables. Those will be added to `nlp.vocab.lookups` from where they can be accessed by components.
- Add JSONL-formatted [vocabulary data](https://spacy.io/api/data-formats#vocab-jsonl) to pre-populate the lexical attributes.
- Load vectors into the pipeline. Vectors are defined as a name or path to a saved `nlp` object containing the vectors, e.g. `en_vectors_web_lg`. It's loaded and the vectors are ported over, while ensuring that all source strings are available in the destination strings. We also warn if there's a mismatch between sourced vectors, since this can lead to problems.
This directory includes additional documentation and explanations of spaCy's internals. It's mostly intended for the spaCy core development team and contributors interested in the more complex parts of the library. The documents generally focus on more abstract implementation details and how specific methods and algorithms work, and they assume knowledge of what's already available in the [usage documentation](https://spacy.io/usage) and [API reference](https://spacy.io/api).
If you're looking to contribute to spaCy, make sure to check out the documentation and [contributing guide](https://github.com/explosion/spaCy/blob/master/CONTRIBUTING.md) first.
This is a list of all the active repos relevant to spaCy besides the main one, with short descriptions, history, and current status. Archived repos will not be covered.
## Always Included in spaCy
These packages are always pulled in when you install spaCy. Most of them are direct dependencies, but some are transitive dependencies through other packages.
- [spacy-legacy](https://github.com/explosion/spacy-legacy): When an architecture in spaCy changes enough to get a new version, the old version is frozen and moved to spacy-legacy. This allows us to keep the core library slim while also preserving backwards compatability.
- [thinc](https://github.com/explosion/thinc): Thinc is the machine learning library that powers trainable components in spaCy. It wraps backends like Numpy, PyTorch, and Tensorflow to provide a functional interface for specifying architectures.
- [catalogue](https://github.com/explosion/catalogue): Small library for adding function registries, like those used for model architectures in spaCy.
- [confection](https://github.com/explosion/confection): This library contains the functionality for config parsing that was formerly contained directly in Thinc.
- [spacy-loggers](https://github.com/explosion/spacy-loggers): Contains loggers beyond the default logger available in spaCy's core code base. This includes loggers integrated with third-party services, which may differ in release cadence from spaCy itself.
- [wasabi](https://github.com/explosion/wasabi): A command line formatting library, used for terminal output in spaCy.
- [srsly](https://github.com/explosion/srsly): A wrapper that vendors several serialization libraries for spaCy. Includes parsers for JSON, JSONL, MessagePack, (extended) Pickle, and YAML.
- [preshed](https://github.com/explosion/preshed): A Cython library for low-level data structures like hash maps, used for memory efficient data storage.
- [cython-blis](https://github.com/explosion/cython-blis): Fast matrix multiplication using BLIS without depending on system libraries. Required by Thinc, rather than spaCy directly.
- [murmurhash](https://github.com/explosion/murmurhash): A wrapper library for a C++ murmurhash implementation, used for string IDs in spaCy and preshed.
- [cymem](https://github.com/explosion/cymem): A small library for RAII-style memory management in Cython.
## Optional Extensions for spaCy
These are repos that can be used by spaCy but aren't part of a default installation. Many of these are wrappers to integrate various kinds of third-party libraries.
- [spacy-transformers](https://github.com/explosion/spacy-transformers): A wrapper for the [HuggingFace Transformers](https://huggingface.co/docs/transformers/index) library, this handles the extensive conversion necessary to coordinate spaCy's powerful `Doc` representation, training pipeline, and the Transformer embeddings. When released, this was known as `spacy-pytorch-transformers`, but it changed to the current name when HuggingFace update the name of their library as well.
- [spacy-huggingface-hub](https://github.com/explosion/spacy-huggingface-hub): This package has a CLI script for uploading a packaged spaCy pipeline (created with `spacy package`) to the [Hugging Face Hub](https://huggingface.co/models).
- [spacy-alignments](https://github.com/explosion/spacy-alignments): A wrapper for the tokenizations library (mentioned below) with a modified build system to simplify cross-platform wheel creation. Used in spacy-transformers for aligning spaCy and HuggingFace tokenizations.
- [spacy-experimental](https://github.com/explosion/spacy-experimental): Experimental components that are not quite ready for inclusion in the main spaCy library. Usually there are unresolved questions around their APIs, so the experimental library allows us to expose them to the community for feedback before fully integrating them.
- [spacy-lookups-data](https://github.com/explosion/spacy-lookups-data): A repository of linguistic data, such as lemmas, that takes up a lot of disk space. Originally created to reduce the size of the spaCy core library. This is mainly useful if you want the data included but aren't using a pretrained pipeline; for the affected languages, the relevant data is included in pretrained pipelines directly.
- [coreferee](https://github.com/explosion/coreferee): Coreference resolution for English, French, German and Polish, optimised for limited training data and easily extensible for further languages. Used as a spaCy pipeline component.
- [spacy-stanza](https://github.com/explosion/spacy-stanza): This is a wrapper that allows the use of Stanford's Stanza library in spaCy.
- [spacy-streamlit](https://github.com/explosion/spacy-streamlit): A wrapper for the Streamlit dashboard building library to help with integrating [displaCy](https://spacy.io/api/top-level/#displacy).
- [spacymoji](https://github.com/explosion/spacymoji): A library to add extra support for emoji to spaCy, such as including character names.
- [thinc-apple-ops](https://github.com/explosion/thinc-apple-ops): A special backend for OSX that uses Apple's native libraries for improved performance.
- [os-signpost](https://github.com/explosion/os-signpost): A Python package that allows you to use the `OSSignposter` API in OSX for performance analysis.
- [spacy-ray](https://github.com/explosion/spacy-ray): A wrapper to integrate spaCy with Ray, a distributed training framework. Currently a work in progress.
## Prodigy
[Prodigy](https://prodi.gy) is Explosion's easy to use and highly customizable tool for annotating data. Prodigy itself requires a license, but the repos below contain documentation, examples, and editor or notebook integrations.
- [prodigy-recipes](https://github.com/explosion/prodigy-recipes): Sample recipes for Prodigy, along with notebooks and other examples of usage.
- [vscode-prodigy](https://github.com/explosion/vscode-prodigy): A VS Code extension that lets you run Prodigy inside VS Code.
- [jupyterlab-prodigy](https://github.com/explosion/jupyterlab-prodigy): An extension for JupyterLab that lets you run Prodigy inside JupyterLab.
## Independent Tools or Projects
These are tools that may be related to or use spaCy, but are functional independent projects in their own right as well.
- [floret](https://github.com/explosion/floret): A modification of fastText to use Bloom Embeddings. Can be used to add vectors with subword features to spaCy, and also works independently in the same manner as fastText.
- [sense2vec](https://github.com/explosion/sense2vec): A library to make embeddings of noun phrases or words coupled with their part of speech. This library uses spaCy.
- [spacy-vectors-builder](https://github.com/explosion/spacy-vectors-builder): This is a spaCy project that builds vectors using floret and a lot of input text. It handles downloading the input data as well as the actual building of vectors.
- [holmes-extractor](https://github.com/explosion/holmes-extractor): Information extraction from English and German texts based on predicate logic. Uses spaCy.
- [healthsea](https://github.com/explosion/healthsea): Healthsea is a project to extract information from comments about health supplements. Structurally, it's a self-contained, large spaCy project.
- [spacy-pkuseg](https://github.com/explosion/spacy-pkuseg): A fork of the pkuseg Chinese tokenizer. Used for Chinese support in spaCy, but also works independently.
- [ml-datasets](https://github.com/explosion/ml-datasets): This repo includes loaders for several standard machine learning datasets, like MNIST or WikiNER, and has historically been used in spaCy example code and documentation.
## Documentation and Informational Repos
These repos are used to support the spaCy docs or otherwise present information about spaCy or other Explosion projects.
- [projects](https://github.com/explosion/projects): The projects repo is used to show detailed examples of spaCy usage. Individual projects can be checked out using the spaCy command line tool, rather than checking out the projects repo directly.
- [spacy-course](https://github.com/explosion/spacy-course): Home to the interactive spaCy course for learning about how to use the library and some basic NLP principles.
- [spacy-io-binder](https://github.com/explosion/spacy-io-binder): Home to the notebooks used for interactive examples in the documentation.
## Organizational / Meta
These repos are used for organizing data around spaCy, but are not something an end user would need to install as part of using the library.
- [spacy-models](https://github.com/explosion/spacy-models): This repo contains metadata (but not training data) for all the spaCy models. This includes information about where their training data came from, version compatability, and performance information. It also includes tests for the model packages, and the built models are hosted as releases of this repo.
- [wheelwright](https://github.com/explosion/wheelwright): A tool for automating our PyPI builds and releases.
- [ec2buildwheel](https://github.com/explosion/ec2buildwheel): A small project that allows you to build Python packages in the manner of cibuildwheel, but on any EC2 image. Used by wheelwright.
## Other
Repos that don't fit in any of the above categories.
- [blis](https://github.com/explosion/blis): A fork of the official BLIS library. The main branch is not updated, but work continues in various branches. This is used for cython-blis.
- [tokenizations](https://github.com/explosion/tokenizations): A library originally by Yohei Tamura to align strings with tolerance to some variations in features like case and diacritics, used for aligning tokens and wordpieces. Adopted and maintained by Explosion, but usually spacy-alignments is used instead.
- [conll-2012](https://github.com/explosion/conll-2012): A repo to hold some slightly cleaned up versions of the official scripts for the CoNLL 2012 shared task involving coreference resolution. Used in the coref project.
- [fastapi-explosion-extras](https://github.com/explosion/fastapi-explosion-extras): Some small tweaks to FastAPI used at Explosion.
spaCy represents mosts strings internally using a `uint64` in Cython which
corresponds to a hash. The magic required to make this largely transparent is
handled by the `StringStore`, and is integrated into the pipelines using the
`Vocab`, which also connects it to some other information.
These are mostly internal details that average library users should never have
to think about. On the other hand, when developing a component it's normal to
interact with the Vocab for lexeme data or word vectors, and it's not unusual
to add labels to the `StringStore`.
## StringStore
### Overview
The `StringStore` is a `cdef class` that looks a bit like a two-way dictionary,
though it is not a subclass of anything in particular.
The main functionality of the `StringStore` is that `__getitem__` converts
hashes into strings or strings into hashes.
The full details of the conversion are complicated. Normally you shouldn't have
to worry about them, but the first applicable case here is used to get the
return value:
1. 0 and the empty string are special cased to each other
2. internal symbols use a lookup table (`SYMBOLS_BY_STR`)
3. normal strings or bytes are hashed
4. internal symbol IDs in `SYMBOLS_BY_INT` are handled
5. anything not yet handled is used as a hash to lookup a string
For the symbol enums, see [`symbols.pxd`](https://github.com/explosion/spaCy/blob/master/spacy/symbols.pxd).
Almost all strings in spaCy are stored in the `StringStore`. This naturally
includes tokens, but also includes things like labels (not just NER/POS/dep,
but also categories etc.), lemmas, lowercase forms, word shapes, and so on. One
of the main results of this is that tokens can be represented by a compact C
struct ([`LexemeC`](https://spacy.io/api/cython-structs#lexemec)/[`TokenC`](https://github.com/explosion/spaCy/issues/4854)) that mostly consists of string hashes. This also means that converting
input for the models is straightforward, and there's not a token mapping step
like in many machine learning frameworks. Additionally, because the token IDs
in spaCy are based on hashes, they are consistent across environments or
models.
One pattern you'll see a lot in spaCy APIs is that `something.value` returns an
`int` and `something.value_` returns a string. That's implemented using the
`StringStore`. Typically the `int` is stored in a C struct and the string is
generated via a property that calls into the `StringStore` with the `int`.
Besides `__getitem__`, the `StringStore` has functions to return specifically a
string or specifically a hash, regardless of whether the input was a string or
hash to begin with, though these are only used occasionally.
### Implementation Details: Hashes and Allocations
Hashes are 64-bit and are computed using [murmurhash][] on UTF-8 bytes. There is no
mechanism for detecting and avoiding collisions. To date there has never been a
reproducible collision or user report about any related issues.
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Matthew Honnibal
