Blinded by Generated Contexts: How Language Models Merge Generated and Retrieved Contexts When Knowledge Conflicts?
- URL: http://arxiv.org/abs/2401.11911v6
- Date: Tue, 11 Jun 2024 02:52:58 GMT
- Title: Blinded by Generated Contexts: How Language Models Merge Generated and Retrieved Contexts When Knowledge Conflicts?
- Authors: Hexiang Tan, Fei Sun, Wanli Yang, Yuanzhuo Wang, Qi Cao, Xueqi Cheng,
- Abstract summary: We identify whether responses are attributed to generated or retrieved contexts.
Experiments reveal a significant bias in several LLMs to favor generated contexts, even when they provide incorrect information.
- Score: 45.233517779029334
- License: http://creativecommons.org/publicdomain/zero/1.0/
- Abstract: While auxiliary information has become a key to enhancing Large Language Models (LLMs), relatively little is known about how LLMs merge these contexts, specifically contexts generated by LLMs and those retrieved from external sources. To investigate this, we formulate a systematic framework to identify whether LLMs' responses are attributed to either generated or retrieved contexts. To easily trace the origin of the response, we construct datasets with conflicting contexts, i.e., each question is paired with both generated and retrieved contexts, yet only one of them contains the correct answer. Our experiments reveal a significant bias in several LLMs (GPT-4/3.5 and Llama2) to favor generated contexts, even when they provide incorrect information. We further identify two key factors contributing to this bias: i) contexts generated by LLMs typically show greater similarity to the questions, increasing their likelihood of being selected; ii) the segmentation process used in retrieved contexts disrupts their completeness, thereby hindering their full utilization in LLMs. Our analysis enhances the understanding of how LLMs merge diverse contexts, offers valuable insights for advancing current LLM augmentation methods, and highlights the risk of generated misinformation for retrieval-augmented LLMs.
Related papers
- Peering into the Mind of Language Models: An Approach for Attribution in Contextual Question Answering [9.86691461253151]
We introduce a novel method for attribution in contextual question answering, leveraging the hidden state representations of large language models (LLMs)
Our approach bypasses the need for extensive model retraining and retrieval model overhead, offering granular attributions and preserving the quality of generated answers.
We present Verifiability-granular, an attribution dataset which has token level annotations for LLM generations in the contextual question answering setup.
arXiv Detail & Related papers (2024-05-28T09:12:44Z) - Unsupervised Information Refinement Training of Large Language Models for Retrieval-Augmented Generation [128.01050030936028]
We propose an information refinement training method named InFO-RAG.
InFO-RAG is low-cost and general across various tasks.
It improves the performance of LLaMA2 by an average of 9.39% relative points.
arXiv Detail & Related papers (2024-02-28T08:24:38Z) - ReSLLM: Large Language Models are Strong Resource Selectors for
Federated Search [35.44746116088232]
Federated search will become increasingly pivotal in the context of Retrieval-Augmented Generation pipelines.
Current SOTA resource selection methodologies rely on feature-based learning approaches.
We propose ReSLLM to drive the selection of resources in federated search in a zero-shot setting.
arXiv Detail & Related papers (2024-01-31T07:58:54Z) - See the Unseen: Better Context-Consistent Knowledge-Editing by Noises [73.54237379082795]
Knowledge-editing updates knowledge of large language models (LLMs)
Existing works ignore this property and the editing lacks generalization.
We empirically find that the effects of different contexts upon LLMs in recalling the same knowledge follow a Gaussian-like distribution.
arXiv Detail & Related papers (2024-01-15T09:09:14Z) - Supervised Knowledge Makes Large Language Models Better In-context Learners [94.89301696512776]
Large Language Models (LLMs) exhibit emerging in-context learning abilities through prompt engineering.
The challenge of improving the generalizability and factuality of LLMs in natural language understanding and question answering remains under-explored.
We propose a framework that enhances the reliability of LLMs as it: 1) generalizes out-of-distribution data, 2) elucidates how LLMs benefit from discriminative models, and 3) minimizes hallucinations in generative tasks.
arXiv Detail & Related papers (2023-12-26T07:24:46Z) - DIVKNOWQA: Assessing the Reasoning Ability of LLMs via Open-Domain
Question Answering over Knowledge Base and Text [73.68051228972024]
Large Language Models (LLMs) have exhibited impressive generation capabilities, but they suffer from hallucinations when relying on their internal knowledge.
Retrieval-augmented LLMs have emerged as a potential solution to ground LLMs in external knowledge.
arXiv Detail & Related papers (2023-10-31T04:37:57Z) - Investigating Answerability of LLMs for Long-Form Question Answering [35.41413072729483]
We focus on long-form question answering (LFQA) because it has several practical and impactful applications.
We propose a question-generation method from abstractive summaries and show that generating follow-up questions from summaries of long documents can create a challenging setting.
arXiv Detail & Related papers (2023-09-15T07:22:56Z) - Check Your Facts and Try Again: Improving Large Language Models with
External Knowledge and Automated Feedback [127.75419038610455]
Large language models (LLMs) are able to generate human-like, fluent responses for many downstream tasks.
This paper proposes a LLM-Augmenter system, which augments a black-box LLM with a set of plug-and-play modules.
arXiv Detail & Related papers (2023-02-24T18:48:43Z)
This list is automatically generated from the titles and abstracts of the papers in this site.
This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.