# Unsupervised Learning of Sentence Embeddings using Compositional n-Gram Features

## TL;DR

This paper introduces an unsupervised method for learning sentence embeddings using compositional n-gram features, which outperforms previous models on various benchmarks, demonstrating robust semantic representations.

## Contribution

It proposes a novel unsupervised objective for training sentence embeddings based on compositional n-gram features, improving upon existing models.

## Key findings

- Outperforms state-of-the-art unsupervised models on benchmark tasks
- Produces robust and general-purpose sentence embeddings
- Demonstrates effectiveness across multiple evaluation metrics

## Abstract

The recent tremendous success of unsupervised word embeddings in a multitude of applications raises the obvious question if similar methods could be derived to improve embeddings (i.e. semantic representations) of word sequences as well. We present a simple but efficient unsupervised objective to train distributed representations of sentences. Our method outperforms the state-of-the-art unsupervised models on most benchmark tasks, highlighting the robustness of the produced general-purpose sentence embeddings.