Package

com.johnsnowlabs.nlp

embeddings

Permalink

package embeddings

Visibility
  1. Public
  2. All

Type Members

  1. abstract class ApproachWithWordEmbeddings[A <: ApproachWithWordEmbeddings[A, M], M <: Model[M] with ModelWithWordEmbeddings] extends AnnotatorApproach[M] with HasEmbeddings

    Permalink

    Base class for annotators that uses Word Embeddings.

    Base class for annotators that uses Word Embeddings. This implementation is based on RocksDB so it has a compact RAM usage

    1. User configures Word Embeddings by method 'setWordEmbeddingsSource'. 2. During training Word Embeddings are indexed as RockDB index file. 3. Than this index file is spread across the cluster. 4. Every model 'ModelWithWordEmbeddings' uses local RocksDB as Word Embeddings lookup.

  2. class ClusterWordEmbeddings extends Serializable

    Permalink
  3. trait EmbeddingsReadable[T <: ModelWithWordEmbeddings] extends ParamsAndFeaturesReadable[T]

    Permalink
  4. trait HasEmbeddings extends AutoCloseable with ParamsAndFeaturesWritable

    Permalink
  5. trait ModelWithWordEmbeddings extends HasEmbeddings

    Permalink

    Base class for models that uses Word Embeddings.

    Base class for models that uses Word Embeddings. This implementation is based on RocksDB so it has a compact RAM usage

    Corresponding Approach have to implement AnnotatorWithWordEmbeddings

  6. case class WordEmbeddingsRetriever(dbFile: String, nDims: Int, caseSensitive: Boolean, lruCacheSize: Int = 100000) extends Closeable with Product with Serializable

    Permalink

Value Members

  1. object ClusterWordEmbeddings extends Serializable

    Permalink
  2. object EmbeddingsHelper

    Permalink
  3. object WordEmbeddingsFormat extends Enumeration

    Permalink
  4. object WordEmbeddingsIndexer

    Permalink

Ungrouped