Package text_embeddings

text_embeddings is a package for no-vocabulary text embeddings.

Expand source code
#!/usr/bin/env python
# -*- coding: utf-8 -*-
# @Date    : 2021-04-17 08:49:42
# @Author  : Chenghao Mou (mouchenghao@gmail.com)

"""text_embeddings is a package for no-vocabulary text embeddings."""

Sub-modules

text_embeddings.base

base covers all the base classes, functions for other embedding based tokenizers.

text_embeddings.byte
text_embeddings.hash

Hash related tokenizers.

text_embeddings.pruning
text_embeddings.visual

Visual information based tokenizers.

text_embeddings.x

X is a Perceiver-based encoder model that incorporates byte hash embeddings, learned token pruning and layer wise adaptive computation (inspired from …