Word Embeddings

This module provides data types for pre-trained word embeddings.

Word embeddings are dense vector representations of words, useful for NLP tasks, semantic similarity, and as input features for neural models.

Base Class

XPM Configdatamaestro_text.data.embeddings.WordEmbeddings(*, id)

Bases: Base

Generic word embeddings

id: str

The unique (sub-)dataset ID

Abstract base class for word embeddings. Provides the load() method that returns a tuple of (words, vectors) where:

• words is a list of vocabulary words

• vectors is a numpy matrix where vectors[i] is the embedding for words[i]

File-Based Embeddings

XPM Configdatamaestro_text.data.embeddings.WordEmbeddingsText(*, id, path, encoding)

Bases: WordEmbeddings, File

Word embeddings as a text word / values

id: str

The unique (sub-)dataset ID

path: path

The path of the file

encoding: str = utf-8

Word embeddings stored in a text file with format: word value1 value2 ... valueN

Example usage:

from datamaestro import prepare_dataset

# Load GloVe embeddings (50-dimensional)
glove = prepare_dataset("edu.stanford.glove.6b.50")

# Load into memory
words, vectors = glove.load()

# Create a word-to-index mapping
word_to_idx = {word: idx for idx, word in enumerate(words)}

# Get embedding for a word
if "computer" in word_to_idx:
    embedding = vectors[word_to_idx["computer"]]
    print(f"Embedding shape: {embedding.shape}")

# Available GloVe variants:
# - edu.stanford.glove.6b.50   (50d, trained on 6B tokens)
# - edu.stanford.glove.6b.100  (100d)
# - edu.stanford.glove.6b.200  (200d)
# - edu.stanford.glove.6b.300  (300d)