SufiSent - Universal Sentence Representations Using Suffix Encodings

TL;DR

SufiSent introduces a novel method for creating universal sentence embeddings by encoding suffixes of word sequences, trained on SNLI, and evaluated on SentEval, showing improved transfer task performance.

Contribution

The paper presents a new suffix encoding technique for sentence representations, enhancing transfer learning in NLP tasks.

Findings

Outperforms existing methods on multiple transfer tasks

Effective on SentEval benchmark

Demonstrates the utility of suffix encodings in sentence embeddings

Abstract

Computing universal distributed representations of sentences is a fundamental task in natural language processing. We propose a method to learn such representations by encoding the suffixes of word sequences in a sentence and training on the Stanford Natural Language Inference (SNLI) dataset. We demonstrate the effectiveness of our approach by evaluating it on the SentEval benchmark, improving on existing approaches on several transfer tasks.