GloVe Embeddings 6B 300 (Multilingual)

Description

GloVe (Global Vectors) is a model for distributed word representation. This is achieved by mapping words into a meaningful space where the distance between words is related to semantic similarity. It outperformed many common Word2vec models on the word analogy task. One benefit of GloVe is that it is the result of directly modeling relationships, instead of getting them as a side effect of training a language model.

Live Demo Open in Colab Download Copy S3 URI

How to use

...
embeddings = WordEmbeddings.pretrained("glove_6B_300", "xx") \
.setInputCols("sentence", "token") \
.setOutputCol("embeddings")
nlp_pipeline = Pipeline(stages=[document_assembler, sentence_detector, tokenizer, embeddings])
pipeline_model = nlp_pipeline.fit(spark.createDataFrame([[""]]).toDF("text"))
result = pipeline_model.transform(spark.createDataFrame([['I love Spark NLP']], ["text"]))

val embeddings = WordEmbeddings.pretrained("glove_6B_300", "xx")
.setInputCols("sentence", "token")
.setOutputCol("embeddings")
val pipeline = new Pipeline().setStages(Array(document_assembler, sentence_detector, tokenizer, embeddings))
val data = Seq("I love Spark NLP").toDF("text")
val result = pipeline.fit(data).transform(data)

import nlu

text = ["""I love Spark NLP"""]
glove_df = nlu.load('xx.embed.glove.6B_300').predict(text)
glove_df

Results

token  |  glove_embeddings                                  |
-------|----------------------------------------------------|
I	| [0.1941000074148178, 0.22603000700473785, -0.4...] |
love	| [0.13948999345302582, 0.534529983997345, -0.25...] |
Spark	| [0.20353999733924866, 0.6292600035667419, 0.27...] |
NLP	| [0.059436000883579254, 0.18411000072956085, -0...] |

Model Information

Data Source

The model is imported from https://nlp.stanford.edu/projects/glove/