Spacy embedding model. The Spacy embedding model used for generating document and word embeddings. Arguments: embedding_model: A spacy embedding model Examples: To create a Spacy backend, you need to create an nlp object and pass it through this backend: ```py
| 5 | |
| 6 | |
| 7 | class SpacyBackend(BaseEmbedder): |
| 8 | """Spacy embedding model. |
| 9 | |
| 10 | The Spacy embedding model used for generating document and |
| 11 | word embeddings. |
| 12 | |
| 13 | Arguments: |
| 14 | embedding_model: A spacy embedding model |
| 15 | |
| 16 | Examples: |
| 17 | To create a Spacy backend, you need to create an nlp object and |
| 18 | pass it through this backend: |
| 19 | |
| 20 | ```python |
| 21 | import spacy |
| 22 | from bertopic.backend import SpacyBackend |
| 23 | |
| 24 | nlp = spacy.load("en_core_web_md", exclude=['tagger', 'parser', 'ner', 'attribute_ruler', 'lemmatizer']) |
| 25 | spacy_model = SpacyBackend(nlp) |
| 26 | ``` |
| 27 | |
| 28 | To load in a transformer model use the following: |
| 29 | |
| 30 | ```python |
| 31 | import spacy |
| 32 | from thinc.api import set_gpu_allocator, require_gpu |
| 33 | from bertopic.backend import SpacyBackend |
| 34 | |
| 35 | nlp = spacy.load("en_core_web_trf", exclude=['tagger', 'parser', 'ner', 'attribute_ruler', 'lemmatizer']) |
| 36 | set_gpu_allocator("pytorch") |
| 37 | require_gpu(0) |
| 38 | spacy_model = SpacyBackend(nlp) |
| 39 | ``` |
| 40 | |
| 41 | If you run into gpu/memory-issues, please use: |
| 42 | |
| 43 | ```python |
| 44 | import spacy |
| 45 | from bertopic.backend import SpacyBackend |
| 46 | |
| 47 | spacy.prefer_gpu() |
| 48 | nlp = spacy.load("en_core_web_trf", exclude=['tagger', 'parser', 'ner', 'attribute_ruler', 'lemmatizer']) |
| 49 | spacy_model = SpacyBackend(nlp) |
| 50 | ``` |
| 51 | """ |
| 52 | |
| 53 | def __init__(self, embedding_model): |
| 54 | super().__init__() |
| 55 | |
| 56 | if "spacy" in str(type(embedding_model)): |
| 57 | self.embedding_model = embedding_model |
| 58 | else: |
| 59 | raise ValueError( |
| 60 | "Please select a correct Spacy model by either using a string such as 'en_core_web_md' " |
| 61 | "or create a nlp model using: `nlp = spacy.load('en_core_web_md')" |
| 62 | ) |
| 63 | |
| 64 | def embed(self, documents: List[str], verbose: bool = False) -> np.ndarray: |