LACoS-BLOOM: Low-rank Adaptation with Contrastive objective on 8 bits Siamese-BLOOM

TL;DR

LACoS-BLOOM introduces a low-rank adaptation with a contrastive objective on 8-bit Siamese-BLOOM, enabling efficient multilingual sentence embeddings with improved performance and reduced resource requirements.

Contribution

The paper presents a novel combination of 8-bit weight quantization, LoRA fine-tuning, and contrastive learning on BLOOM for multilingual sentence similarity tasks.

Findings

Embeddings quality improves with more parameters and data.

Achieves better results than Sentence-BERT on STS tasks.

Runs efficiently on a single GPU with 7.1B parameters.

Abstract

Text embeddings are useful features for several NLP applications, such as sentence similarity, text clustering, and semantic search. In this paper, we present a Low-rank Adaptation with a Contrastive objective on top of 8-bit Siamese-BLOOM, a multilingual large language model optimized to produce semantically meaningful word embeddings. The innovation is threefold. First, we cast BLOOM weights to 8-bit values. Second, we fine-tune BLOOM with a scalable adapter (LoRA) and 8-bit Adam optimizer for sentence similarity classification. Third, we apply a Siamese architecture on BLOOM model with a contrastive objective to ease the multi-lingual labeled data scarcity. The experiment results show the quality of learned embeddings from LACoS-BLOOM is proportional to the number of model parameters and the amount of unlabeled training data. With the parameter efficient fine-tuning design, we are able to run BLOOM 7.1 billion parameters end-to-end on a single GPU machine with 32GB memory. Compared to previous solution Sentence-BERT, we achieve significant improvement on both English and multi-lingual STS tasks.