DoLa: Decoding by Contrasting Layers Improves Factuality in Large Language Models

By Yung-Sung Chuang, Yujia Xie, Hongyin Luo, Yoon Kim, James R. Glass, Pengcheng He. International Conference on Learning Representations, 2023

397 citations recorded by the source corpus.

Despite their impressive capabilities, large language models (LLMs) are prone to hallucinations, i.e., generating content that deviates from facts seen during pretraining. We propose a simple decoding strategy for reducing hallucinations with pretrained LLMs that does not require conditioning on retrieved external knowledge nor additional fine-tuning. Our approach obtains the next-token distribution by contrasting the differences in logits obtained from projecting the later layers versus earlier layers to the vocabulary space, exploiting the fact that factual knowledge in an LLMs has generally been shown to be localized to particular transformer layers. We find that this Decoding by Contrasting Layers (DoLa) approach is able to better surface factual knowledge and reduce the generation of incorrect facts. DoLa consistently improves the truthfulness across multiple choices tasks and open-ended generation tasks, for example improving the performance of LLaMA family models on TruthfulQA by 12-17% absolute points, demonstrating its potential in making LLMs reliably generate truthful facts.

Subfields: Artificial Intelligence

View Paper: https://www.semanticscholar.org/paper/ed5020eeda1fbe8c29b1282d654b34abee22d90f

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Satyapriya Krishna
2. Himabindu Lakkaraju
3. Chirag Agarwal
4. Tong Wang

Citations in the corpus

• OPERA: Alleviating Hallucination in Multi-Modal Large Language Models via Over-Trust Penalty and Retrospection-Allocation (2023, neutral): “DoLa [8] decoding is a recently proposed decoding method that aims to mitigate the hallucinations in MLLMs, which contrasts the logits of mature layer and pre-mature layers and rescale the increments as the output.”
• Understanding the Effects of Iterative Prompting on Truthfulness (2024, background): “Furthermore, innovative decoding strategies that contrast different layers within a model have proven effective in reducing inaccuracies and hallucinations, contributing to the overall factual consistency of model-generated content [19].”
• Hallucination of Multimodal Large Language Models: A Survey (2024, neutral): “The lack of standardized benchmarks and evaluation metrics poses significant challenges in assessing the degree of hallucination in MLLMs.”
• Siren's Song in the AI Ocean: A Survey on Hallucination in Large Language Models (2023, neutral): “Another work, DoLA (Chuang et al., 2023), also employ the idea of contrastive decoding to reduce hallucination.”