spaCy/spacy/vectors.pyx

688 lines
25 KiB
Cython
Raw Normal View History

cimport numpy as np
from libc.stdint cimport uint32_t, uint64_t
from cython.operator cimport dereference as deref
from libcpp.set cimport set as cppset
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
from murmurhash.mrmr cimport hash128_x64
import functools
2017-06-05 10:32:08 +00:00
import numpy
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
from typing import cast
import warnings
from enum import Enum
import srsly
from thinc.api import Ops, get_array_module, get_current_ops
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
from thinc.backends import get_array_ops
from thinc.types import Floats2d
2017-06-05 10:32:08 +00:00
from .strings cimport StringStore
from .strings import get_string_id
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
from .errors import Errors, Warnings
2017-10-27 17:45:19 +00:00
from . import util
2017-06-05 10:32:08 +00:00
2018-03-10 21:53:42 +00:00
def unpickle_vectors(bytes_data):
return Vectors().from_bytes(bytes_data)
2017-10-31 17:25:08 +00:00
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
class Mode(str, Enum):
default = "default"
floret = "floret"
@classmethod
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
def values(cls):
return list(cls.__members__.keys())
2017-06-05 10:32:08 +00:00
cdef class Vectors:
2017-10-27 17:45:19 +00:00
"""Store, save and load word vectors.
2017-10-01 22:05:54 +00:00
2017-10-01 20:10:33 +00:00
Vectors data is kept in the vectors.data attribute, which should be an
2017-10-27 17:45:19 +00:00
instance of numpy.ndarray (for CPU vectors) or cupy.ndarray
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
(for GPU vectors).
In the default mode, `vectors.key2row` is a dictionary mapping word hashes
to rows in the vectors.data table. Multiple keys can be mapped to the same
vector, and not all of the rows in the table need to be assigned - so
len(list(vectors.keys())) may be greater or smaller than vectors.shape[0].
2017-10-31 22:23:34 +00:00
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
In floret mode, the floret settings (minn, maxn, etc.) are used to
calculate the vector from the rows corresponding to the key's ngrams.
DOCS: https://spacy.io/api/vectors
2017-10-27 17:45:19 +00:00
"""
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
cdef public object strings
cdef public object name
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
cdef readonly object mode
2017-06-05 10:32:08 +00:00
cdef public object data
2017-08-19 02:33:03 +00:00
cdef public object key2row
cdef cppset[int] _unset
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
cdef readonly uint32_t minn
cdef readonly uint32_t maxn
cdef readonly uint32_t hash_count
cdef readonly uint32_t hash_seed
cdef readonly unicode bow
cdef readonly unicode eow
def __init__(self, *, strings=None, shape=None, data=None, keys=None, name=None, mode=Mode.default, minn=0, maxn=0, hash_count=1, hash_seed=0, bow="<", eow=">"):
2017-10-31 17:25:08 +00:00
"""Create a new vector store.
2017-10-31 22:23:34 +00:00
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
strings (StringStore): The string store.
2017-10-31 17:25:08 +00:00
shape (tuple): Size of the table, as (# entries, # columns)
data (numpy.ndarray or cupy.ndarray): The vector data.
2017-10-31 22:23:34 +00:00
keys (iterable): A sequence of keys, aligned with the data.
2020-05-24 15:20:58 +00:00
name (str): A name to identify the vectors table.
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
mode (str): Vectors mode: "default" or "floret" (default: "default").
minn (int): The floret char ngram minn (default: 0).
maxn (int): The floret char ngram maxn (default: 0).
hash_count (int): The floret hash count (1-4, default: 1).
hash_seed (int): The floret hash seed (default: 0).
bow (str): The floret BOW string (default: "<").
eow (str): The floret EOW string (default: ">").
DOCS: https://spacy.io/api/vectors#init
2017-10-27 17:45:19 +00:00
"""
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
self.strings = strings
if self.strings is None:
self.strings = StringStore()
self.name = name
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
if mode not in Mode.values():
raise ValueError(
Errors.E202.format(
name="vectors",
mode=mode,
modes=str(Mode.values())
)
)
self.mode = Mode(mode).value
self.key2row = {}
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
self.minn = minn
self.maxn = maxn
self.hash_count = hash_count
self.hash_seed = hash_seed
self.bow = bow
self.eow = eow
if self.mode == Mode.default:
if data is None:
if shape is None:
shape = (0,0)
ops = get_current_ops()
data = ops.xp.zeros(shape, dtype="f")
self._unset = cppset[int]({i for i in range(data.shape[0])})
else:
self._unset = cppset[int]()
self.data = data
if keys is not None:
for i, key in enumerate(keys):
self.add(key, row=i)
elif self.mode == Mode.floret:
if maxn < minn:
raise ValueError(Errors.E863)
if hash_count < 1 or hash_count >= 5:
raise ValueError(Errors.E862)
if data is None:
raise ValueError(Errors.E864)
if keys is not None:
raise ValueError(Errors.E861)
self.data = data
self._unset = cppset[int]()
2017-10-31 22:23:34 +00:00
2017-10-31 17:25:08 +00:00
@property
def shape(self):
"""Get `(rows, dims)` tuples of number of rows and number of dimensions
in the vector table.
RETURNS (tuple): A `(rows, dims)` pair.
DOCS: https://spacy.io/api/vectors#shape
2017-10-31 17:25:08 +00:00
"""
return self.data.shape
@property
def size(self):
"""The vector size i,e. rows * dims.
RETURNS (int): The vector size.
DOCS: https://spacy.io/api/vectors#size
"""
return self.data.size
2017-10-31 17:25:08 +00:00
@property
def is_full(self):
"""Whether the vectors table is full.
RETURNS (bool): `True` if no slots are available for new keys.
DOCS: https://spacy.io/api/vectors#is_full
"""
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
if self.mode == Mode.floret:
return True
return self._unset.size() == 0
2017-06-05 10:32:08 +00:00
2017-10-31 18:30:52 +00:00
@property
def n_keys(self):
"""Get the number of keys in the table. Note that this is the number
of all keys, not just unique vectors.
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
RETURNS (int): The number of keys in the table for default vectors.
For floret vectors, return -1.
DOCS: https://spacy.io/api/vectors#n_keys
"""
if self.mode == Mode.floret:
return -1
2017-10-31 18:30:52 +00:00
return len(self.key2row)
2017-06-05 10:32:08 +00:00
def __reduce__(self):
2018-03-10 21:53:42 +00:00
return (unpickle_vectors, (self.to_bytes(),))
2017-06-05 10:32:08 +00:00
def __getitem__(self, key):
2017-10-31 17:25:08 +00:00
"""Get a vector by key. If the key is not found, a KeyError is raised.
2017-10-01 20:10:33 +00:00
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
key (str/int): The key to get the vector for.
2017-10-31 17:25:08 +00:00
RETURNS (ndarray): The vector for the key.
DOCS: https://spacy.io/api/vectors#getitem
2017-10-27 17:45:19 +00:00
"""
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
if self.mode == Mode.default:
i = self.key2row.get(get_string_id(key), None)
if i is None:
raise KeyError(Errors.E058.format(key=key))
else:
return self.data[i]
elif self.mode == Mode.floret:
return self.get_batch([key])[0]
raise KeyError(Errors.E058.format(key=key))
2017-06-05 10:32:08 +00:00
def __setitem__(self, key, vector):
2017-10-31 17:25:08 +00:00
"""Set a vector for the given key.
2017-10-27 17:45:19 +00:00
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
key (str/int): The key to set the vector for.
2017-10-31 22:23:34 +00:00
vector (ndarray): The vector to set.
DOCS: https://spacy.io/api/vectors#setitem
2017-10-27 17:45:19 +00:00
"""
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
if self.mode == Mode.floret:
warnings.warn(Warnings.W115.format(method="Vectors.__setitem__"))
return
key = get_string_id(key)
2017-08-19 02:33:03 +00:00
i = self.key2row[key]
2017-06-05 10:32:08 +00:00
self.data[i] = vector
if self._unset.count(i):
self._unset.erase(self._unset.find(i))
2017-06-05 10:32:08 +00:00
def __iter__(self):
2017-10-31 22:23:34 +00:00
"""Iterate over the keys in the table.
2017-10-27 17:45:19 +00:00
2017-10-31 22:23:34 +00:00
YIELDS (int): A key in the table.
DOCS: https://spacy.io/api/vectors#iter
2017-10-27 17:45:19 +00:00
"""
2017-10-31 17:25:08 +00:00
yield from self.key2row
2017-06-05 10:32:08 +00:00
def __len__(self):
2017-10-31 17:25:08 +00:00
"""Return the number of vectors in the table.
2017-10-27 17:45:19 +00:00
RETURNS (int): The number of vectors in the data.
DOCS: https://spacy.io/api/vectors#len
2017-10-27 17:45:19 +00:00
"""
2017-10-31 17:25:08 +00:00
return self.data.shape[0]
2017-08-19 17:52:25 +00:00
def __contains__(self, key):
2017-10-31 17:25:08 +00:00
"""Check whether a key has been mapped to a vector entry in the table.
2017-10-27 17:45:19 +00:00
2017-10-31 17:25:08 +00:00
key (int): The key to check.
2017-10-27 17:45:19 +00:00
RETURNS (bool): Whether the key has a vector entry.
DOCS: https://spacy.io/api/vectors#contains
2017-10-27 17:45:19 +00:00
"""
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
if self.mode == Mode.floret:
return True
else:
return key in self.key2row
2017-08-19 17:52:25 +00:00
2017-10-31 17:25:08 +00:00
def resize(self, shape, inplace=False):
2017-10-31 22:23:34 +00:00
"""Resize the underlying vectors array. If inplace=True, the memory
2017-10-31 17:25:08 +00:00
is reallocated. This may cause other references to the data to become
invalid, so only use inplace=True if you're sure that's what you want.
If the number of vectors is reduced, keys mapped to rows that have been
deleted are removed. These removed items are returned as a list of
2017-10-31 22:23:34 +00:00
`(key, row)` tuples.
shape (tuple): A `(rows, dims)` tuple.
inplace (bool): Reallocate the memory.
RETURNS (list): The removed items as a list of `(key, row)` tuples.
DOCS: https://spacy.io/api/vectors#resize
2017-10-31 22:23:34 +00:00
"""
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
if self.mode == Mode.floret:
warnings.warn(Warnings.W115.format(method="Vectors.resize"))
return -1
xp = get_array_module(self.data)
2017-10-31 17:25:08 +00:00
if inplace:
if shape[1] != self.data.shape[1]:
raise ValueError(Errors.E193.format(new_dim=shape[1], curr_dim=self.data.shape[1]))
if xp == numpy:
self.data.resize(shape, refcheck=False)
else:
raise ValueError(Errors.E192)
2017-10-31 17:25:08 +00:00
else:
resized_array = xp.zeros(shape, dtype=self.data.dtype)
copy_shape = (min(shape[0], self.data.shape[0]), min(shape[1], self.data.shape[1]))
resized_array[:copy_shape[0], :copy_shape[1]] = self.data[:copy_shape[0], :copy_shape[1]]
self.data = resized_array
self._sync_unset()
2017-10-31 17:25:08 +00:00
removed_items = []
for key, row in self.key2row.copy().items():
2017-10-31 17:25:08 +00:00
if row >= shape[0]:
self.key2row.pop(key)
removed_items.append((key, row))
return removed_items
2017-10-31 22:23:34 +00:00
2017-10-31 17:25:08 +00:00
def keys(self):
"""RETURNS (iterable): A sequence of keys in the table."""
2017-10-31 22:23:34 +00:00
return self.key2row.keys()
2017-10-31 17:25:08 +00:00
def values(self):
2017-10-31 22:23:34 +00:00
"""Iterate over vectors that have been assigned to at least one key.
2017-10-31 17:25:08 +00:00
Note that some vectors may be unassigned, so the number of vectors
2017-10-31 22:23:34 +00:00
returned may be less than the length of the vectors table.
YIELDS (ndarray): A vector in the table.
DOCS: https://spacy.io/api/vectors#values
2017-10-31 22:23:34 +00:00
"""
2017-10-31 17:25:08 +00:00
for row, vector in enumerate(range(self.data.shape[0])):
if not self._unset.count(row):
2017-10-31 17:25:08 +00:00
yield vector
def items(self):
"""Iterate over `(key, vector)` pairs.
YIELDS (tuple): A key/vector pair.
DOCS: https://spacy.io/api/vectors#items
2017-10-31 17:25:08 +00:00
"""
for key, row in self.key2row.items():
yield key, self.data[row]
2017-10-31 23:34:55 +00:00
def find(self, *, key=None, keys=None, row=None, rows=None):
2017-10-31 23:42:39 +00:00
"""Look up one or more keys by row, or vice versa.
2017-10-31 23:34:55 +00:00
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
key (Union[int, str]): Find the row that the given key points to.
2017-10-31 23:34:55 +00:00
Returns int, -1 if missing.
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
keys (Iterable[Union[int, str]]): Find rows that the keys point to.
2017-10-31 23:34:55 +00:00
Returns ndarray.
2019-03-16 16:10:57 +00:00
row (int): Find the first key that points to the row.
2017-10-31 23:34:55 +00:00
Returns int.
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
rows (Iterable[int]): Find the keys that point to the rows.
2017-10-31 23:34:55 +00:00
Returns ndarray.
2017-10-31 23:42:39 +00:00
RETURNS: The requested key, keys, row or rows.
"""
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
if self.mode == Mode.floret:
raise ValueError(
Errors.E858.format(
mode=self.mode,
alternative="Use Vectors[key] instead.",
)
)
2017-10-31 23:34:55 +00:00
if sum(arg is None for arg in (key, keys, row, rows)) != 3:
bad_kwargs = {"key": key, "keys": keys, "row": row, "rows": rows}
raise ValueError(Errors.E059.format(kwargs=bad_kwargs))
2017-10-31 17:25:08 +00:00
xp = get_array_module(self.data)
2017-10-31 23:34:55 +00:00
if key is not None:
key = get_string_id(key)
2017-10-31 23:34:55 +00:00
return self.key2row.get(key, -1)
elif keys is not None:
keys = [get_string_id(key) for key in keys]
rows = [self.key2row.get(key, -1) for key in keys]
return xp.asarray(rows, dtype="i")
2017-10-31 23:34:55 +00:00
else:
row2key = {row: key for key, row in self.key2row.items()}
2017-10-31 23:34:55 +00:00
if row is not None:
return row2key[row]
2017-10-31 23:34:55 +00:00
else:
results = [row2key[row] for row in rows]
return xp.asarray(results, dtype="uint64")
2017-10-31 17:25:08 +00:00
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
def _get_ngram_hashes(self, unicode s):
"""Calculate up to 4 32-bit hash values with MurmurHash3_x64_128 using
the floret hash settings.
key (str): The string key.
RETURNS: A list of the integer hashes.
"""
# MurmurHash3_x64_128 returns an array of 2 uint64_t values.
cdef uint64_t[2] out
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
chars = s.encode("utf8")
cdef char* utf8_string = chars
hash128_x64(utf8_string, len(chars), self.hash_seed, &out)
rows = [
out[0] & 0xffffffffu,
out[0] >> 32,
out[1] & 0xffffffffu,
out[1] >> 32,
]
return rows[:min(self.hash_count, 4)]
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
def _get_ngrams(self, unicode key):
"""Get all padded ngram strings using the ngram settings.
key (str): The string key.
RETURNS: A list of the ngram strings for the padded key.
"""
key = self.bow + key + self.eow
ngrams = [key] + [
key[start:start+ngram_size]
for ngram_size in range(self.minn, self.maxn + 1)
for start in range(0, len(key) - ngram_size + 1)
]
return ngrams
def get_batch(self, keys):
"""Get the vectors for the provided keys efficiently as a batch.
keys (Iterable[Union[int, str]]): The keys.
RETURNS: The requested vectors from the vector table.
"""
ops = get_array_ops(self.data)
if self.mode == Mode.default:
rows = self.find(keys=keys)
vecs = self.data[rows]
elif self.mode == Mode.floret:
keys = [self.strings.as_string(key) for key in keys]
if sum(len(key) for key in keys) == 0:
return ops.xp.zeros((len(keys), self.data.shape[1]))
unique_keys = tuple(set(keys))
row_index = {key: i for i, key in enumerate(unique_keys)}
rows = [row_index[key] for key in keys]
indices = []
lengths = []
for key in unique_keys:
if key == "":
ngram_rows = []
else:
ngram_rows = [
h % self.data.shape[0]
for ngram in self._get_ngrams(key)
for h in self._get_ngram_hashes(ngram)
]
indices.extend(ngram_rows)
lengths.append(len(ngram_rows))
indices = ops.asarray(indices, dtype="int32")
lengths = ops.asarray(lengths, dtype="int32")
vecs = ops.reduce_mean(cast(Floats2d, self.data[indices]), lengths)
vecs = vecs[rows]
return ops.as_contig(vecs)
def add(self, key, *, vector=None, row=None):
"""Add a key to the table. Keys can be mapped to an existing vector
by setting `row`, or a new vector can be added.
2017-10-27 17:45:19 +00:00
key (int): The key to add.
vector (ndarray / None): A vector to add for the key.
row (int / None): The row number of a vector to map the key to.
RETURNS (int): The row the vector was added to.
DOCS: https://spacy.io/api/vectors#add
2017-10-27 17:45:19 +00:00
"""
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
if self.mode == Mode.floret:
warnings.warn(Warnings.W115.format(method="Vectors.add"))
return -1
# use int for all keys and rows in key2row for more efficient access
# and serialization
key = int(get_string_id(key))
if row is not None:
row = int(row)
2017-10-31 01:00:26 +00:00
if row is None and key in self.key2row:
row = self.key2row[key]
elif row is None:
2017-10-31 17:25:08 +00:00
if self.is_full:
raise ValueError(Errors.E060.format(rows=self.data.shape[0],
cols=self.data.shape[1]))
row = deref(self._unset.begin())
if row < self.data.shape[0]:
self.key2row[key] = row
else:
raise ValueError(Errors.E197.format(row=row, key=key))
2017-08-19 17:52:25 +00:00
if vector is not None:
xp = get_array_module(self.data)
vector = xp.asarray(vector)
self.data[row] = vector
if self._unset.count(row):
self._unset.erase(self._unset.find(row))
return row
2017-10-31 22:23:34 +00:00
def most_similar(self, queries, *, batch_size=1024, n=1, sort=True):
"""For each of the given vectors, find the n most similar entries
2017-10-31 17:25:08 +00:00
to it, by cosine.
2017-10-31 22:23:34 +00:00
Queries are by vector. Results are returned as a `(keys, best_rows,
scores)` tuple. If `queries` is large, the calculations are performed in
chunks, to avoid consuming too much memory. You can set the `batch_size`
to control the size/space trade-off during the calculations.
queries (ndarray): An array with one or more vectors.
batch_size (int): The batch size to use.
n (int): The number of entries to return for each query.
sort (bool): Whether to sort the n entries returned by score.
RETURNS (tuple): The most similar entries as a `(keys, best_rows, scores)`
tuple.
"""
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
if self.mode == Mode.floret:
raise ValueError(Errors.E858.format(
mode=self.mode,
alternative="",
))
xp = get_array_module(self.data)
filled = sorted(list({row for row in self.key2row.values()}))
if len(filled) < n:
raise ValueError(Errors.E198.format(n=n, n_rows=len(filled)))
filled = xp.asarray(filled)
2017-10-31 22:23:34 +00:00
norms = xp.linalg.norm(self.data[filled], axis=1, keepdims=True)
norms[norms == 0] = 1
vectors = self.data[filled] / norms
2017-10-31 22:23:34 +00:00
best_rows = xp.zeros((queries.shape[0], n), dtype='i')
scores = xp.zeros((queries.shape[0], n), dtype='f')
2017-10-31 17:25:08 +00:00
# Work in batches, to avoid memory problems.
for i in range(0, queries.shape[0], batch_size):
batch = queries[i : i+batch_size]
batch_norms = xp.linalg.norm(batch, axis=1, keepdims=True)
batch_norms[batch_norms == 0] = 1
batch /= batch_norms
2017-10-31 17:25:08 +00:00
# batch e.g. (1024, 300)
# vectors e.g. (10000, 300)
# sims e.g. (1024, 10000)
sims = xp.dot(batch, vectors.T)
best_rows[i:i+batch_size] = xp.argpartition(sims, -n, axis=1)[:,-n:]
scores[i:i+batch_size] = xp.partition(sims, -n, axis=1)[:,-n:]
if sort and n >= 2:
sorted_index = xp.arange(scores.shape[0])[:,None][i:i+batch_size],xp.argsort(scores[i:i+batch_size], axis=1)[:,::-1]
scores[i:i+batch_size] = scores[sorted_index]
best_rows[i:i+batch_size] = best_rows[sorted_index]
for i, j in numpy.ndindex(best_rows.shape):
best_rows[i, j] = filled[best_rows[i, j]]
# Round values really close to 1 or -1
scores = xp.around(scores, decimals=4, out=scores)
# Account for numerical error we want to return in range -1, 1
scores = xp.clip(scores, a_min=-1, a_max=1, out=scores)
2017-11-01 01:06:58 +00:00
row2key = {row: key for key, row in self.key2row.items()}
numpy_rows = get_current_ops().to_numpy(best_rows)
keys = xp.asarray(
[[row2key[row] for row in numpy_rows[i] if row in row2key]
for i in range(len(queries)) ], dtype="uint64")
return (keys, best_rows, scores)
2017-06-05 10:32:08 +00:00
def to_ops(self, ops: Ops):
self.data = ops.asarray(self.data)
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
def _get_cfg(self):
if self.mode == Mode.default:
return {
"mode": Mode(self.mode).value,
}
elif self.mode == Mode.floret:
return {
"mode": Mode(self.mode).value,
"minn": self.minn,
"maxn": self.maxn,
"hash_count": self.hash_count,
"hash_seed": self.hash_seed,
"bow": self.bow,
"eow": self.eow,
}
def _set_cfg(self, cfg):
self.mode = Mode(cfg.get("mode", Mode.default)).value
self.minn = cfg.get("minn", 0)
self.maxn = cfg.get("maxn", 0)
self.hash_count = cfg.get("hash_count", 0)
self.hash_seed = cfg.get("hash_seed", 0)
self.bow = cfg.get("bow", "<")
self.eow = cfg.get("eow", ">")
def to_disk(self, path, *, exclude=tuple()):
2017-10-27 17:45:19 +00:00
"""Save the current state to a directory.
2020-05-24 15:20:58 +00:00
path (str / Path): A path to a directory, which will be created if
it doesn't exists.
DOCS: https://spacy.io/api/vectors#to_disk
2017-10-27 17:45:19 +00:00
"""
xp = get_array_module(self.data)
if xp is numpy:
save_array = lambda arr, file_: xp.save(file_, arr, allow_pickle=False)
else:
save_array = lambda arr, file_: xp.save(file_, arr)
def save_vectors(path):
# the source of numpy.save indicates that the file object is closed after use.
# but it seems that somehow this does not happen, as ResourceWarnings are raised here.
# in order to not rely on this, wrap in context manager.
ops = get_current_ops()
with path.open("wb") as _file:
save_array(ops.to_numpy(self.data, byte_order="<"), _file)
serializers = {
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
"strings": lambda p: self.strings.to_disk(p.with_suffix(".json")),
2020-06-20 13:52:00 +00:00
"vectors": lambda p: save_vectors(p),
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
"key2row": lambda p: srsly.write_msgpack(p, self.key2row),
"vectors.cfg": lambda p: srsly.write_json(p, self._get_cfg()),
}
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
return util.to_disk(path, serializers, exclude)
2017-08-18 18:45:48 +00:00
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
def from_disk(self, path, *, exclude=tuple()):
2017-10-27 17:45:19 +00:00
"""Loads state from a directory. Modifies the object in place and
returns it.
2020-05-24 15:20:58 +00:00
path (str / Path): Directory path, string or Path-like object.
2017-10-27 17:45:19 +00:00
RETURNS (Vectors): The modified object.
DOCS: https://spacy.io/api/vectors#from_disk
2017-10-27 17:45:19 +00:00
"""
2017-10-31 18:58:35 +00:00
def load_key2row(path):
2017-08-19 20:07:00 +00:00
if path.exists():
self.key2row = srsly.read_msgpack(path)
2017-10-31 18:58:35 +00:00
for key, row in self.key2row.items():
if self._unset.count(row):
self._unset.erase(self._unset.find(row))
2017-10-31 18:58:35 +00:00
def load_keys(path):
if path.exists():
keys = numpy.load(str(path))
for i, key in enumerate(keys):
self.add(key, row=i)
2017-08-19 16:42:11 +00:00
def load_vectors(path):
Update spaCy for thinc 8.0.0 (#4920) * Add load_from_config function * Add train_from_config script * Merge configs and expose via spacy.config * Fix script * Suggest create_evaluation_callback * Hard-code for NER * Fix errors * Register command * Add TODO * Update train-from-config todos * Fix imports * Allow delayed setting of parser model nr_class * Get train-from-config working * Tidy up and fix scores and printing * Hide traceback if cancelled * Fix weighted score formatting * Fix score formatting * Make output_path optional * Add Tok2Vec component * Tidy up and add tok2vec_tensors * Add option to copy docs in nlp.update * Copy docs in nlp.update * Adjust nlp.update() for set_annotations * Don't shuffle pipes in nlp.update, decruft * Support set_annotations arg in component update * Support set_annotations in parser update * Add get_gradients method * Add get_gradients to parser * Update errors.py * Fix problems caused by merge * Add _link_components method in nlp * Add concept of 'listeners' and ControlledModel * Support optional attributes arg in ControlledModel * Try having tok2vec component in pipeline * Fix tok2vec component * Fix config * Fix tok2vec * Update for Example * Update for Example * Update config * Add eg2doc util * Update and add schemas/types * Update schemas * Fix nlp.update * Fix tagger * Remove hacks from train-from-config * Remove hard-coded config str * Calculate loss in tok2vec component * Tidy up and use function signatures instead of models * Support union types for registry models * Minor cleaning in Language.update * Make ControlledModel specifically Tok2VecListener * Fix train_from_config * Fix tok2vec * Tidy up * Add function for bilstm tok2vec * Fix type * Fix syntax * Fix pytorch optimizer * Add example configs * Update for thinc describe changes * Update for Thinc changes * Update for dropout/sgd changes * Update for dropout/sgd changes * Unhack gradient update * Work on refactoring _ml * Remove _ml.py module * WIP upgrade cli scripts for thinc * Move some _ml stuff to util * Import link_vectors from util * Update train_from_config * Import from util * Import from util * Temporarily add ml.component_models module * Move ml methods * Move typedefs * Update load vectors * Update gitignore * Move imports * Add PrecomputableAffine * Fix imports * Fix imports * Fix imports * Fix missing imports * Update CLI scripts * Update spacy.language * Add stubs for building the models * Update model definition * Update create_default_optimizer * Fix import * Fix comment * Update imports in tests * Update imports in spacy.cli * Fix import * fix obsolete thinc imports * update srsly pin * from thinc to ml_datasets for example data such as imdb * update ml_datasets pin * using STATE.vectors * small fix * fix Sentencizer.pipe * black formatting * rename Affine to Linear as in thinc * set validate explicitely to True * rename with_square_sequences to with_list2padded * rename with_flatten to with_list2array * chaining layernorm * small fixes * revert Optimizer import * build_nel_encoder with new thinc style * fixes using model's get and set methods * Tok2Vec in component models, various fixes * fix up legacy tok2vec code * add model initialize calls * add in build_tagger_model * small fixes * setting model dims * fixes for ParserModel * various small fixes * initialize thinc Models * fixes * consistent naming of window_size * fixes, removing set_dropout * work around Iterable issue * remove legacy tok2vec * util fix * fix forward function of tok2vec listener * more fixes * trying to fix PrecomputableAffine (not succesful yet) * alloc instead of allocate * add morphologizer * rename residual * rename fixes * Fix predict function * Update parser and parser model * fixing few more tests * Fix precomputable affine * Update component model * Update parser model * Move backprop padding to own function, for test * Update test * Fix p. affine * Update NEL * build_bow_text_classifier and extract_ngrams * Fix parser init * Fix test add label * add build_simple_cnn_text_classifier * Fix parser init * Set gpu off by default in example * Fix tok2vec listener * Fix parser model * Small fixes * small fix for PyTorchLSTM parameters * revert my_compounding hack (iterable fixed now) * fix biLSTM * Fix uniqued * PyTorchRNNWrapper fix * small fixes * use helper function to calculate cosine loss * small fixes for build_simple_cnn_text_classifier * putting dropout default at 0.0 to ensure the layer gets built * using thinc util's set_dropout_rate * moving layer normalization inside of maxout definition to optimize dropout * temp debugging in NEL * fixed NEL model by using init defaults ! * fixing after set_dropout_rate refactor * proper fix * fix test_update_doc after refactoring optimizers in thinc * Add CharacterEmbed layer * Construct tagger Model * Add missing import * Remove unused stuff * Work on textcat * fix test (again :)) after optimizer refactor * fixes to allow reading Tagger from_disk without overwriting dimensions * don't build the tok2vec prematuraly * fix CharachterEmbed init * CharacterEmbed fixes * Fix CharacterEmbed architecture * fix imports * renames from latest thinc update * one more rename * add initialize calls where appropriate * fix parser initialization * Update Thinc version * Fix errors, auto-format and tidy up imports * Fix validation * fix if bias is cupy array * revert for now * ensure it's a numpy array before running bp in ParserStepModel * no reason to call require_gpu twice * use CupyOps.to_numpy instead of cupy directly * fix initialize of ParserModel * remove unnecessary import * fixes for CosineDistance * fix device renaming * use refactored loss functions (Thinc PR 251) * overfitting test for tagger * experimental settings for the tagger: avoid zero-init and subword normalization * clean up tagger overfitting test * use previous default value for nP * remove toy config * bringing layernorm back (had a bug - fixed in thinc) * revert setting nP explicitly * remove setting default in constructor * restore values as they used to be * add overfitting test for NER * add overfitting test for dep parser * add overfitting test for textcat * fixing init for linear (previously affine) * larger eps window for textcat * ensure doc is not None * Require newer thinc * Make float check vaguer * Slop the textcat overfit test more * Fix textcat test * Fix exclusive classes for textcat * fix after renaming of alloc methods * fixing renames and mandatory arguments (staticvectors WIP) * upgrade to thinc==8.0.0.dev3 * refer to vocab.vectors directly instead of its name * rename alpha to learn_rate * adding hashembed and staticvectors dropout * upgrade to thinc 8.0.0.dev4 * add name back to avoid warning W020 * thinc dev4 * update srsly * using thinc 8.0.0a0 ! Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com> Co-authored-by: Ines Montani <ines@ines.io>
2020-01-29 16:06:46 +00:00
ops = get_current_ops()
2017-08-19 20:07:00 +00:00
if path.exists():
Update spaCy for thinc 8.0.0 (#4920) * Add load_from_config function * Add train_from_config script * Merge configs and expose via spacy.config * Fix script * Suggest create_evaluation_callback * Hard-code for NER * Fix errors * Register command * Add TODO * Update train-from-config todos * Fix imports * Allow delayed setting of parser model nr_class * Get train-from-config working * Tidy up and fix scores and printing * Hide traceback if cancelled * Fix weighted score formatting * Fix score formatting * Make output_path optional * Add Tok2Vec component * Tidy up and add tok2vec_tensors * Add option to copy docs in nlp.update * Copy docs in nlp.update * Adjust nlp.update() for set_annotations * Don't shuffle pipes in nlp.update, decruft * Support set_annotations arg in component update * Support set_annotations in parser update * Add get_gradients method * Add get_gradients to parser * Update errors.py * Fix problems caused by merge * Add _link_components method in nlp * Add concept of 'listeners' and ControlledModel * Support optional attributes arg in ControlledModel * Try having tok2vec component in pipeline * Fix tok2vec component * Fix config * Fix tok2vec * Update for Example * Update for Example * Update config * Add eg2doc util * Update and add schemas/types * Update schemas * Fix nlp.update * Fix tagger * Remove hacks from train-from-config * Remove hard-coded config str * Calculate loss in tok2vec component * Tidy up and use function signatures instead of models * Support union types for registry models * Minor cleaning in Language.update * Make ControlledModel specifically Tok2VecListener * Fix train_from_config * Fix tok2vec * Tidy up * Add function for bilstm tok2vec * Fix type * Fix syntax * Fix pytorch optimizer * Add example configs * Update for thinc describe changes * Update for Thinc changes * Update for dropout/sgd changes * Update for dropout/sgd changes * Unhack gradient update * Work on refactoring _ml * Remove _ml.py module * WIP upgrade cli scripts for thinc * Move some _ml stuff to util * Import link_vectors from util * Update train_from_config * Import from util * Import from util * Temporarily add ml.component_models module * Move ml methods * Move typedefs * Update load vectors * Update gitignore * Move imports * Add PrecomputableAffine * Fix imports * Fix imports * Fix imports * Fix missing imports * Update CLI scripts * Update spacy.language * Add stubs for building the models * Update model definition * Update create_default_optimizer * Fix import * Fix comment * Update imports in tests * Update imports in spacy.cli * Fix import * fix obsolete thinc imports * update srsly pin * from thinc to ml_datasets for example data such as imdb * update ml_datasets pin * using STATE.vectors * small fix * fix Sentencizer.pipe * black formatting * rename Affine to Linear as in thinc * set validate explicitely to True * rename with_square_sequences to with_list2padded * rename with_flatten to with_list2array * chaining layernorm * small fixes * revert Optimizer import * build_nel_encoder with new thinc style * fixes using model's get and set methods * Tok2Vec in component models, various fixes * fix up legacy tok2vec code * add model initialize calls * add in build_tagger_model * small fixes * setting model dims * fixes for ParserModel * various small fixes * initialize thinc Models * fixes * consistent naming of window_size * fixes, removing set_dropout * work around Iterable issue * remove legacy tok2vec * util fix * fix forward function of tok2vec listener * more fixes * trying to fix PrecomputableAffine (not succesful yet) * alloc instead of allocate * add morphologizer * rename residual * rename fixes * Fix predict function * Update parser and parser model * fixing few more tests * Fix precomputable affine * Update component model * Update parser model * Move backprop padding to own function, for test * Update test * Fix p. affine * Update NEL * build_bow_text_classifier and extract_ngrams * Fix parser init * Fix test add label * add build_simple_cnn_text_classifier * Fix parser init * Set gpu off by default in example * Fix tok2vec listener * Fix parser model * Small fixes * small fix for PyTorchLSTM parameters * revert my_compounding hack (iterable fixed now) * fix biLSTM * Fix uniqued * PyTorchRNNWrapper fix * small fixes * use helper function to calculate cosine loss * small fixes for build_simple_cnn_text_classifier * putting dropout default at 0.0 to ensure the layer gets built * using thinc util's set_dropout_rate * moving layer normalization inside of maxout definition to optimize dropout * temp debugging in NEL * fixed NEL model by using init defaults ! * fixing after set_dropout_rate refactor * proper fix * fix test_update_doc after refactoring optimizers in thinc * Add CharacterEmbed layer * Construct tagger Model * Add missing import * Remove unused stuff * Work on textcat * fix test (again :)) after optimizer refactor * fixes to allow reading Tagger from_disk without overwriting dimensions * don't build the tok2vec prematuraly * fix CharachterEmbed init * CharacterEmbed fixes * Fix CharacterEmbed architecture * fix imports * renames from latest thinc update * one more rename * add initialize calls where appropriate * fix parser initialization * Update Thinc version * Fix errors, auto-format and tidy up imports * Fix validation * fix if bias is cupy array * revert for now * ensure it's a numpy array before running bp in ParserStepModel * no reason to call require_gpu twice * use CupyOps.to_numpy instead of cupy directly * fix initialize of ParserModel * remove unnecessary import * fixes for CosineDistance * fix device renaming * use refactored loss functions (Thinc PR 251) * overfitting test for tagger * experimental settings for the tagger: avoid zero-init and subword normalization * clean up tagger overfitting test * use previous default value for nP * remove toy config * bringing layernorm back (had a bug - fixed in thinc) * revert setting nP explicitly * remove setting default in constructor * restore values as they used to be * add overfitting test for NER * add overfitting test for dep parser * add overfitting test for textcat * fixing init for linear (previously affine) * larger eps window for textcat * ensure doc is not None * Require newer thinc * Make float check vaguer * Slop the textcat overfit test more * Fix textcat test * Fix exclusive classes for textcat * fix after renaming of alloc methods * fixing renames and mandatory arguments (staticvectors WIP) * upgrade to thinc==8.0.0.dev3 * refer to vocab.vectors directly instead of its name * rename alpha to learn_rate * adding hashembed and staticvectors dropout * upgrade to thinc 8.0.0.dev4 * add name back to avoid warning W020 * thinc dev4 * update srsly * using thinc 8.0.0a0 ! Co-authored-by: Matthew Honnibal <honnibal+gh@gmail.com> Co-authored-by: Ines Montani <ines@ines.io>
2020-01-29 16:06:46 +00:00
self.data = ops.xp.load(str(path))
self.to_ops(ops)
2017-08-18 18:45:48 +00:00
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
def load_settings(path):
if path.exists():
self._set_cfg(srsly.read_json(path))
serializers = {
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
"strings": lambda p: self.strings.from_disk(p.with_suffix(".json")),
"vectors": load_vectors,
2020-06-20 13:52:00 +00:00
"keys": load_keys,
"key2row": load_key2row,
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
"vectors.cfg": load_settings,
}
2017-08-18 18:45:48 +00:00
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
util.from_disk(path, serializers, exclude)
self._sync_unset()
2017-08-19 16:42:11 +00:00
return self
2017-06-05 10:32:08 +00:00
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
def to_bytes(self, *, exclude=tuple()):
2017-10-27 17:45:19 +00:00
"""Serialize the current state to a binary string.
exclude (list): String names of serialization fields to exclude.
2017-10-27 17:45:19 +00:00
RETURNS (bytes): The serialized form of the `Vectors` object.
DOCS: https://spacy.io/api/vectors#to_bytes
2017-10-27 17:45:19 +00:00
"""
2017-06-05 10:32:08 +00:00
def serialize_weights():
if hasattr(self.data, "to_bytes"):
2017-08-18 18:45:48 +00:00
return self.data.to_bytes()
2017-06-05 10:32:08 +00:00
else:
ops = get_current_ops()
return srsly.msgpack_dumps(ops.to_numpy(self.data, byte_order="<"))
serializers = {
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
"strings": lambda: self.strings.to_bytes(),
"key2row": lambda: srsly.msgpack_dumps(self.key2row),
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
"vectors": serialize_weights,
"vectors.cfg": lambda: srsly.json_dumps(self._get_cfg()),
}
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
return util.to_bytes(serializers, exclude)
2017-06-05 10:32:08 +00:00
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
def from_bytes(self, data, *, exclude=tuple()):
2017-10-27 17:45:19 +00:00
"""Load state from a binary string.
data (bytes): The data to load from.
exclude (list): String names of serialization fields to exclude.
2017-10-27 17:45:19 +00:00
RETURNS (Vectors): The `Vectors` object.
DOCS: https://spacy.io/api/vectors#from_bytes
2017-10-27 17:45:19 +00:00
"""
2017-06-05 10:32:08 +00:00
def deserialize_weights(b):
if hasattr(self.data, "from_bytes"):
2017-08-18 18:45:48 +00:00
self.data.from_bytes()
2017-06-05 10:32:08 +00:00
else:
xp = get_array_module(self.data)
self.data = xp.asarray(srsly.msgpack_loads(b))
ops = get_current_ops()
self.to_ops(ops)
2017-06-05 10:32:08 +00:00
deserializers = {
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
"strings": lambda b: self.strings.from_bytes(b),
"key2row": lambda b: self.key2row.update(srsly.msgpack_loads(b)),
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
"vectors": deserialize_weights,
"vectors.cfg": lambda b: self._set_cfg(srsly.json_loads(b))
}
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
util.from_bytes(data, deserializers, exclude)
self._sync_unset()
2017-08-19 16:42:11 +00:00
return self
Add support for floret vectors (#8909) * 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
2021-10-27 12:08:31 +00:00
def clear(self):
"""Clear all entries in the vector table.
DOCS: https://spacy.io/api/vectors#clear
"""
if self.mode == Mode.floret:
raise ValueError(Errors.E859)
self.key2row = {}
self._sync_unset()
def _sync_unset(self):
filled = {row for row in self.key2row.values()}
self._unset = cppset[int]({row for row in range(self.data.shape[0]) if row not in filled})