Lower Layers Matter: Alleviating Hallucination via Multi-Layer Fusion Contrastive Decoding with Truthfulness Refocused
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866916773316526080 |
|---|---|
| author | Chen, Dingwei Fang, Feiteng Ni, Shiwen Liang, Feng Hu, Xiping Argha, Ahmadreza Alinejad-Rokny, Hamid Yang, Min Li, Chengming |
| author_facet | Chen, Dingwei Fang, Feiteng Ni, Shiwen Liang, Feng Hu, Xiping Argha, Ahmadreza Alinejad-Rokny, Hamid Yang, Min Li, Chengming |
| contents | Large Language Models (LLMs) have demonstrated exceptional performance across various natural language processing tasks. However, they occasionally generate inaccurate and counterfactual outputs, a phenomenon commonly referred to as "hallucinations''. To tackle this issue, recent studies have explored contrastive decoding between the original model and an amateur model with induced hallucination, showing promising results. Nevertheless, this approach can disrupt the original LLM's output distribution due to coarse contrast and simple subtraction operations, potentially leading to errors. In this paper, we introduce a novel contrastive decoding framework, termed LOL (LOwer Layer Matters). Unlike prior methods that focus solely on the final layer, our approach integrates contrastive information from lower layers to enable multi-layer fusion during contrastive decoding. Additionally, we incorporate a truthfulness refocused module that leverages instruction guidance to further improve truthfulness in contrastive decoding. Extensive experiments on four publicly available datasets demonstrate that the LOL framework significantly mitigates hallucination while outperforming existing baselines in most cases. For reproducibility, we will release our code and data upon acceptance. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2408_08769 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Lower Layers Matter: Alleviating Hallucination via Multi-Layer Fusion Contrastive Decoding with Truthfulness Refocused Chen, Dingwei Fang, Feiteng Ni, Shiwen Liang, Feng Hu, Xiping Argha, Ahmadreza Alinejad-Rokny, Hamid Yang, Min Li, Chengming Computation and Language Large Language Models (LLMs) have demonstrated exceptional performance across various natural language processing tasks. However, they occasionally generate inaccurate and counterfactual outputs, a phenomenon commonly referred to as "hallucinations''. To tackle this issue, recent studies have explored contrastive decoding between the original model and an amateur model with induced hallucination, showing promising results. Nevertheless, this approach can disrupt the original LLM's output distribution due to coarse contrast and simple subtraction operations, potentially leading to errors. In this paper, we introduce a novel contrastive decoding framework, termed LOL (LOwer Layer Matters). Unlike prior methods that focus solely on the final layer, our approach integrates contrastive information from lower layers to enable multi-layer fusion during contrastive decoding. Additionally, we incorporate a truthfulness refocused module that leverages instruction guidance to further improve truthfulness in contrastive decoding. Extensive experiments on four publicly available datasets demonstrate that the LOL framework significantly mitigates hallucination while outperforming existing baselines in most cases. For reproducibility, we will release our code and data upon acceptance. |
| title | Lower Layers Matter: Alleviating Hallucination via Multi-Layer Fusion Contrastive Decoding with Truthfulness Refocused |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2408.08769 |