Example of embeddings using ELMo on TensorFlow Hub

This repository is an example of word embeddings using ELMo (Embeddings from Language Models) on TensorFlow Hub.

The comparison_test module can calculate cosine similarity between given two sentences.

Usage

• In case of using exmaple.sh

sh example.sh

• In case of executing on your terminal

python comparison_test.py [sentence1] [sentence2]

Example

bash-3.2$ python comparison_test.py "people read the book" "the book people read"

[Cosine Similarity]
"people read the book" vs "the book people read"
ELMo: 0.83875865
NNLM: 1.0