relevanceai.operations.vector.base#

Module Contents#

relevanceai.operations.vector.base.BASE_2VEC_DEFINITON#
class relevanceai.operations.vector.base.Base2Vec#

Base class for vector

classmethod validate_model_url(cls, model_url: str, list_of_urls: List[str])#

Validate the model url belongs in the list of urls. This is to help users to avoid mis-spelling the name of the model.

# TODO: Improve model URL validation to not include final number in URl string.

Parameters
  • model_url – The URl of the the model in question

  • list_of_urls – The list of URLS for the model in question

static is_url_working(url)#
classmethod chunk(self, lst: List, chunksize: int)#

Chunk an iterable object in Python but not a pandas DataFrame. :param lst: Python List :param chunksize: The chunk size of an object.

Example

>>> documents = [{...}]
>>> ViClient.chunk(documents)
property zero_vector(self)#
is_empty_vector(self, vector)#
get_default_vector_field_name(self, field, field_type='vector')#
encode_documents(self, fields: list, documents: list, vector_error_treatment='zero_vector', field_type='vector')#

Encode documents and their specific fields. Note that this runs off the default encode method. If there is a specific function that you want run, ensure that it is set to the encode function.

Parameters
  • missing_treatment – Missing treatment can be one of [“do_not_include”, “zero_vector”, value].

  • documents – The documents that are being used

  • fields – The list of fields to be used

  • field_type – Accepts “vector” or “chunkvector”

encode_chunk_documents(self, chunk_field, fields: list, documents: list, vector_error_treatment: str = 'zero_vector')#

Encode chunk documents. Loops through every field and then every document.

Parameters
  • chunk_field – The field for chunking

  • fields – A list of fields for chunk documents

  • documents – a list of documents

  • vector_error_treatment – Vector Error Treatment

Example

>>> chunk_docs = enc.encode_chunk_documents(chunk_field="value", fields=["text"], documents=chunk_docs)
bulk_encode_documents(self, fields: list, documents: list, vector_error_treatment='zero_vector', field_type='vector')#

Encode documents and their specific fields. Note that this runs off the default encode method. If there is a specific function that you want run, ensure that it is set to the encode function.

Parameters
  • missing_treatment – Missing treatment can be one of [“do_not_include”, “zero_vector”, value].

  • documents – The documents that are being used

  • fields – The list of fields to be used

get_combinations(self, lst, maximum_span=5)#
get_cosine_similarity(self, vector_1, vector_2)#
get_word_combinations(self, sentence, maximum_span=5)#
get_result(self, result_text, query_vector)#
explain(self, query_text, result_texts, highlight_threshold=0.5, max_highlights=None, return_cos_similarity_docs: bool = True, ignore_warnings: bool = True)#