This page is still being polished. If you have thoughts, please share them via the feedback form.
Data on this page is preliminary and may change. Please do not share or cite these figures publicly.
Non-Model mitigations not clearly fitting above categories.
Also in Non-Model
LLM hallucinations caused by the absence of certain domain-specific data can be mitigated through the supplementation of training data. However, in practice, encompassing all conceivable domains within the training corpus is challenging. Therefore, a prevalent approach to mitigating hallucinations is to integrate external knowledge as supporting evidence for content generation. Generally, the external knowledge is utilized as a part of the input [122], [178]–[184] or used as evidence for a post-hoc revision process [271]–[276]. To obtain the external knowledge, pioneer studies retrieve factual triplets from reliable knowledge bases (KBs) [277]–[279].
Nevertheless, KBs typically have limited general knowledge, primarily due to the high cost of human annotations. Hence, information retrieval (IR) systems are used to retrieve evidence from open-ended Web sources (e.g., Wikipedia) [178]. However, information gathered from the Web sources carries noisy information and redundancy, which can mislead LLMs to generate unsatisfied responses. To mitigate this issue, recent endeavors refine models’ responses through automated feedback [178], [280] or clarifications from human users [183]. Besides obtaining external knowledge from aforementioned non-parametric sources, a Parametric Knowledge Guiding (PKG) framework [281] is proposed to use a trainable task-specific module to generate relevant context as the augmented knowledge.
Reasoning
Retrieve external knowledge from information sources to augment model inputs and filter outputs with factual evidence.
Mitigation in Input Modules
Mitigating the threat posed by the input module presents a significant challenge for LLM developers due to the diversity of the harmful inputs and adversarial prompts [209], [210].
1.2.1 Guardrails & FilteringMitigation in Input Modules > Defensive Prompt Design
Directly modifying the input prompts is a viable approach to steer the behavior of the model and foster the generation of responsible outputs. This method integrates contextual information or constraints in the prompts to provide background knowledge and guidelines while generating the output [22].
1.2.1 Guardrails & FilteringMitigation in Input Modules > Malicious Prompt Detection
Different from the methods of designing defensive prompts to preprocess the input, the malicious prompt detection method aims to detect and filter out the harmful prompts through the input safeguard
1.2.1 Guardrails & FilteringMitigation in Language Models
This section delves into mitigating risks associated with models, encompassing privacy preservation, detoxification and debiasing, mitigation of hallucinations, and defenses against model attacks.
1.1 ModelMitigation in Language Models > Privacy Preserving
Privacy leakage is a crucial risk of LLMs, since the powerful memorization and association capabilities of LLMs raise the risk of revealing private information within the training data. Researchers are devoted to designing privacypreserving frameworks in LLMs [226], [227], aiming to safeguard sensitive PII from possible disclosure during humanmachine conservation
1.1 ModelMitigation in Language Models > Detoxifying and Debiasing
To reduce the toxicity and bias of LLMs, prior efforts mainly focus on enhancing the quality of training data and conducting safety training.
1.1 ModelRisk Taxonomy, Mitigation, and Assessment Benchmarks of Large Language Model Systems
Cui, Tianyu; Wang, Yanling; Fu, Chuanpu; Xiao, Yong; Li, Sijia; Deng, Xinhao; Liu, Yunpeng; Zhang, Qinglin; Qiu, Ziyi; Li, Peiyang; Tan, Zhixing; Xiong, Junwu; Kong, Xinyu; Wen, Zujie; Xu, Ke; Li, Qi (2024)
Despite their impressive capabilities, large lan- guage models (LLMs) have been observed to generate responses that include inaccurate or fabricated information, a phenomenon com- monly known as “hallucination”. In this work, we propose a simple Induce-then-Contrast De- coding (ICD) strategy to alleviate hallucina- tions. We first construct a factually weak LLM by inducing hallucinations from the original LLMs. Then, we penalize these induced hallu- cinations during decoding to enhance the fac- tuality of the generated content. Concretely, we determine the final next-token predictions by amplifying the predictions from the orig- inal model and downplaying the induced un- truthful predictions via contrastive decoding. Experimental results on both discrimination- based and generation-based hallucination eval- uation benchmarks, such as TruthfulQA and FACTSCORE, demonstrate that our proposed ICD methods can effectively enhance the factu- ality of LLMs across various model sizes and families. For example, when equipped with ICD, Llama2-7B-Chat and Mistral-7B-Instruct achieve performance comparable to ChatGPT and GPT4 on TruthfulQA, respectively.
Build and Use Model
Training, fine-tuning, and integrating the AI model
Deployer
Entity that integrates and deploys the AI system for end users
Manage
Prioritising, responding to, and mitigating AI risks
