OCEAN: Offline Chain-of-thought Evaluation and Alignment in Large Language Models

• URL: http://arxiv.org/abs/2410.23703v1
• Date: Thu, 31 Oct 2024 07:48:44 GMT
• Title: OCEAN: Offline Chain-of-thought Evaluation and Alignment in Large Language Models
• Authors: Junda Wu, Xintong Li, Ruoyu Wang, Yu Xia, Yuxin Xiong, Jianing Wang, Tong Yu, Xiang Chen, Branislav Kveton, Lina Yao, Jingbo Shang, Julian McAuley,
• Abstract summary: This work focuses on the offline evaluation of the chain-of-thought capabilities of LLMs. We use knowledge graphs (e.g., Wikidata5m) to provide feedback on the generated chain of thoughts. We show how to optimize LLMs based on the proposed evaluation method.
• Score: 68.17018458283651
• License:
• Abstract: Offline evaluation of LLMs is crucial in understanding their capacities, though current methods remain underexplored in existing research. In this work, we focus on the offline evaluation of the chain-of-thought capabilities and show how to optimize LLMs based on the proposed evaluation method. To enable offline feedback with rich knowledge and reasoning paths, we use knowledge graphs (e.g., Wikidata5m) to provide feedback on the generated chain of thoughts. Due to the heterogeneity between LLM reasoning and KG structures, direct interaction and feedback from KGs on LLM behavior are challenging, as they require accurate entity linking and grounding of LLM-generated chains of thought in the KG. To address the above challenge, we propose an offline chain-of-thought evaluation framework, OCEAN, which models chain-of-thought reasoning in LLMs as an MDP and evaluate the policy's alignment with KG preference modeling. To overcome the reasoning heterogeneity and grounding problems, we leverage on-policy KG exploration and RL to model a KG policy that generates token-level likelihood distributions for LLM-generated chain-of-thought reasoning paths, simulating KG reasoning preference. Then we incorporate the knowledge-graph feedback on the validity and alignment of the generated reasoning paths into inverse propensity scores and propose KG-IPS estimator. Theoretically, we prove the unbiasedness of the proposed KG-IPS estimator and provide a lower bound on its variance. With the off-policy evaluated value function, we can directly enable off-policy optimization to further enhance chain-of-thought alignment. Our empirical study shows that OCEAN can be efficiently optimized for generating chain-of-thought reasoning paths with higher estimated values without affecting LLMs' general abilities in downstream tasks or their internal knowledge.

Related papers

• Online Preference Alignment for Language Models via Count-based Exploration [46.46627519343809]
  Reinforcement Learning from Human Feedback (RLHF) has shown great potential in fine-tuning Large Language Models (LLMs) to align with human preferences. Existing methods perform preference alignment from a fixed dataset, which can be limited in data coverage. Online RLHF is more desirable to empower the LLM to explore outside the support of the initial dataset by iteratively collecting the prompt-response pairs.
  arXiv  Detail & Related papers  (2025-01-22T09:12:09Z)
• SelfPrompt: Autonomously Evaluating LLM Robustness via Domain-Constrained Knowledge Guidelines and Refined Adversarial Prompts [0.6291443816903801]
  This paper introduces a novel framework designed to autonomously evaluate the robustness of large language models (LLMs) Our method generates descriptive sentences from domain-constrained knowledge graph triplets to formulate adversarial prompts. This self-evaluation mechanism allows the LLM to evaluate its robustness without the need for external benchmarks.
  arXiv  Detail & Related papers  (2024-12-01T10:58:53Z)
• Simple Is Effective: The Roles of Graphs and Large Language Models in Knowledge-Graph-Based Retrieval-Augmented Generation [9.844598565914055]
  Large Language Models (LLMs) demonstrate strong reasoning abilities but face limitations such as hallucinations and outdated knowledge. We introduce SubgraphRAG, extending the Knowledge Graph (KG)-based Retrieval-Augmented Generation (RAG) framework that retrieves subgraphs. Our approach innovatively integrates a lightweight multilayer perceptron with a parallel triple-scoring mechanism for efficient and flexible subgraph retrieval.
  arXiv  Detail & Related papers  (2024-10-28T04:39:32Z)
• Decoding on Graphs: Faithful and Sound Reasoning on Knowledge Graphs through Generation of Well-Formed Chains [66.55612528039894]
  Knowledge Graphs (KGs) can serve as reliable knowledge sources for question answering (QA) We present DoG (Decoding on Graphs), a novel framework that facilitates a deep synergy between LLMs and KGs. Experiments across various KGQA tasks with different background KGs demonstrate that DoG achieves superior and robust performance.
  arXiv  Detail & Related papers  (2024-10-24T04:01:40Z)
• Can Knowledge Graphs Make Large Language Models More Trustworthy? An Empirical Study Over Open-ended Question Answering [30.12049172634714]
  This study explores whether Knowledge Graphs can make Large Language Models (LLMs) more trustworthy in an open-ended setting. OKGQA is a benchmark specifically designed to assess LLMs enhanced with Knowledge Graphs under open-ended, real-world question answering scenarios. OKGQA-P is a benchmark variant to assess model performance when the semantics and structure of KGs are deliberately perturbed and contaminated.
  arXiv  Detail & Related papers  (2024-10-10T16:29:21Z)
• VinePPO: Unlocking RL Potential For LLM Reasoning Through Refined Credit Assignment [66.80143024475635]
  We propose VinePPO, a straightforward approach to compute unbiased Monte Carlo-based estimates. We show that VinePPO consistently outperforms PPO and other RL-free baselines across MATH and GSM8K datasets.
  arXiv  Detail & Related papers  (2024-10-02T15:49:30Z)
• How Can LLM Guide RL? A Value-Based Approach [68.55316627400683]
  Reinforcement learning (RL) has become the de facto standard practice for sequential decision-making problems by improving future acting policies with feedback. Recent developments in large language models (LLMs) have showcased impressive capabilities in language understanding and generation, yet they fall short in exploration and self-improvement capabilities. We develop an algorithm named LINVIT that incorporates LLM guidance as a regularization factor in value-based RL, leading to significant reductions in the amount of data needed for learning.
  arXiv  Detail & Related papers  (2024-02-25T20:07:13Z)
• Direct Evaluation of Chain-of-Thought in Multi-hop Reasoning with Knowledge Graphs [52.42505579545893]
  Large language models (LLMs) demonstrate strong reasoning abilities when prompted to generate chain-of-thought explanations alongside answers. We propose a novel discriminative and generative CoT evaluation paradigm to assess LLMs' knowledge of reasoning and the accuracy of the generated CoT.
  arXiv  Detail & Related papers  (2024-02-17T05:22:56Z)
• Mitigating Large Language Model Hallucinations via Autonomous Knowledge Graph-based Retrofitting [51.7049140329611]
  This paper proposes Knowledge Graph-based Retrofitting (KGR) to mitigate factual hallucination during the reasoning process. Experiments show that KGR can significantly improve the performance of LLMs on factual QA benchmarks.
  arXiv  Detail & Related papers  (2023-11-22T11:08:38Z)

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.