Pith. sign in

REVIEW 4 major objections 6 minor 8 cited by

Think More, Hallucinate Less: Mitigating Hallucinations via Dual Process of Fast and Slow Thinking

T0 review · 4 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read HaluSearch turns answer generation into a sentence-scored tree search, and reports the best accuracy among compared methods on all six factuality datasets.

desk verdict A useful engineering idea undercut by a circular evaluation and a missing full-system comparison; the headline claim is not supported as stated. read the letter →

arxiv 2501.01306 v2 pith:MHAKHLD3 submitted 2025-01-02 cs.CL

Xiaoxue Cheng , Junyi Li , Wayne Xin Zhao , Ji-Rong Wen This is my paper · ORCID
classification cs.CL
keywords hallucinationmitigationMonteCarlotreesearchinference-timereasoningSystem2thinkingrewardmodelself-evaluationfactualitybenchmarksdualprocesstheory
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Hallucinations in large language models often persist even when the model knows the correct fact; the paper argues this is because ordinary decoding moves too fast and lets one bad sentence contaminate the rest. HaluSearch treats response generation as a step-by-step reasoning process: at each sentence, a Monte Carlo tree search expands several candidate continuations, a reward model scores how likely each continuation is to hallucinate, and the search advances along the most reliable path. To keep this affordable, a learned switch decides when a question or a single step needs slow thinking and when fast direct generation is enough. If the central claim is right, the same model can become more truthful at inference time without retraining, and users can dial the compute-versus-accuracy trade-off. The paper reports gains over prompt-based and inference-time baselines on all six English and Chinese factuality datasets.

What carries the argument

The central object is a Monte Carlo tree in which each node is one generated sentence and each child is an alternative next sentence sampled from the policy model. Selection uses the UCT score, which balances a node's accumulated value against how rarely it has been visited; expansion samples candidate continuations; evaluation completes each candidate with rollouts and averages reward-model scores over them; backpropagation updates ancestor values. The reward model is trained on the same base model, scoring hallucination likelihood on a 1-to-5 scale, with a variant that produces a critique before the score. A separate switch model, trained on labels derived from the search tree's own node values, governs when to think fast and generate one sentence directly versus think slow and expand multiple candidates, at both the level of the whole question and the level of each step.

What would settle it

Take a factuality benchmark and split it by whether the policy model can answer a question when given a one-sentence hint: the paper's internal-knowledge story predicts HaluSearch's accuracy gain over direct generation is concentrated in questions the model can answer with a hint but not without. If the gain is just as large on hint-unanswerable questions, the search is not simply exploiting latent knowledge, and the central premise fails.

Watch

Extended reading notes

Core claim

On the paper's own terms, hallucination mitigation is better posed as a search problem than as a prompting problem. HaluSearch frames text generation not as one pass from prompt to answer but as a tree of possible sentence-level continuations, with every node scored by a reward model trained to predict how likely a continuation is to hallucinate. The claim is that this step-level search fully exploits the policy model's internal knowledge: because the model often knows facts it fails to state, branching and scoring give the reliable path a chance to surface. In the reported experiments, HaluSearch outperforms direct generation, chain-of-thought prompting, self-consistency, best-of-N, self-refinement, and activation shifting on all six English and Chinese datasets. A self-trained reward model that produces a critique before a score is shown to be competitive with a much larger external reward model, supporting the paper's case that no closed-source scorer is required. The intended conclusion is that deliberate, search-based decoding can remove a substantial share of hallucinations at inference time.

Load-bearing premise

The policy model already holds the facts needed to answer the test questions, so that exploring its own generations can surface the correct path; if the knowledge is absent, tree search cannot create it.

Editorial extensions

If this is right

  • Because rewards are assigned per sentence, a wrong intermediate step can be detected and abandoned before it contaminates the rest of the response, reducing error accumulation.
  • The trained critique-based reward model makes the approach deployable without a closed-source scorer, since it matches a larger external reward model on TruthfulQA.
  • The switch threshold gives an explicit accuracy-efficiency dial: at the highest threshold tested, average solving time per question falls from 53.3 to 25.4 seconds while accuracy drops from 42.6% to 37.6% on HaluEval-QA.
  • The gains hold across languages and model families: both Llama3.1-8B-Instruct and Qwen2-7B-Instruct improve on all six English and Chinese benchmarks.
  • Scaling the number of expanded nodes and rollouts improves accuracy up to a point, after which the policy model's internal knowledge and the reward model's scoring accuracy become the limiting factors.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • One consequence the authors leave implicit: the same sentence-level search should transfer to long-form generation, since any multi-sentence passage can be decomposed and scored the same way.
  • The paper reports accuracy and time separately; a direct cost-matched comparison against Best-of-N at equal total generated tokens would show whether the advantage comes from the search structure itself or from the larger number of generated candidates, an open question.
  • A natural untested extension is to close the loop by retraining the reward and switch models on data produced by HaluSearch itself, which could compound the gains shown in the single training round reported.
  • The dual-process design suggests a general resource-allocation principle for inference-time compute: spend extra computation where the model's own uncertainty is high and skip it elsewhere, a principle that could be implemented in decoding algorithms other than tree search.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The paper proposes HaluSearch, an inference-time framework that uses Monte Carlo Tree Search (MCTS) to generate responses sentence by sentence, with a reward model scoring each step and a trained switch model deciding between fast (System 1) and slow (System 2) thinking at both instance and step levels. The self-evaluation reward model is trained on GPT-4-generated scores and critiques, and the full method is referred to as MCTSwitch. The authors evaluate on six English and Chinese hallucination datasets with Llama3.1-8B-Instruct and Qwen2-7B-Instruct as policy models, reporting that MCTS outperforms direct generation, Chain-of-Thought, Self-Consistency, Best-of-N, Self-Refine, and ITI in Table 1. The paper also analyzes trained reward models and the efficiency-accuracy trade-off of the switch mechanism, and concludes that HaluSearch significantly mitigates hallucinations.

Significance. The idea of applying step-level tree search to hallucination mitigation is timely and the framework is clearly described: the MCTS formulation, the reward model training procedure, and the switch mechanism are all specified in sufficient detail to be reproduced. If the empirical claim is established with an evaluation that is independent of the optimization signal, the paper would be a useful contribution to inference-time hallucination mitigation and to the growing literature on System 2 reasoning in LLMs. The paper also provides concrete ablations of reward model variants (Generative RM and Generative RM + Critic) and of the switch threshold, which are informative. However, the current experimental validation has a central weakness: the same model, GPT-4, is used both as the reward signal that guides the search and as the judge that scores the final answers, so the headline gains in Table 1 may partly reflect reward hacking rather than genuine factuality improvements.

major comments (4)
  1. [§4.1, Section 3.2, Table 1] The evaluation is not independent of the optimization target. In Implementation Details, GPT-4 is used as the reward model inside MCTS, and in Datasets and Metrics, GPT-4 is also used to judge the correctness of every generated response. Since the search explicitly selects sentences and rollouts that maximize GPT-4's hallucination score, the reported accuracy is measured by the same model whose preferences are being optimized. Baselines such as CoT, Self-Consistency, and Best-of-N do not receive this GPT-4 steering, so the comparison is asymmetric. The trained self-evaluation reward models in Section 4.3 are distilled from GPT-4 labels and therefore inherit the same bias. Please re-evaluate with a judge that is not used anywhere in reward training or search, or with human evaluation, and report the results. Without this, the claim in Section 4.2 that HaluSearch 'achieves the best performance across all six Chinese and English datasets' is not supported.
  2. [§4.2, Table 1, §4.4, Figure 2] The main table reports a row labeled 'MCTS', not the full HaluSearch pipeline with the dynamic switch (MCTSwitch). The switch mechanism is a stated contribution and is evaluated only on HaluEval-QA in Section 4.4, with a threshold-based analysis. Moreover, the internal numbers are inconsistent: Figure 2 reports 42.6% accuracy for 100% slow thinking on HaluEval-QA with Llama3.1-8B-Instruct, which is lower than the 45.4% MCTS row in Table 1 for the same policy model, dataset, and reward setup. This discrepancy needs to be explained, and the full MCTSwitch system should be compared against the baselines on all six datasets, or the paper should explicitly state which configuration corresponds to the headline claim.
  3. [§4.4, Algorithm 1] The switch threshold analysis appears to be conducted on the same dataset used to create the switch training data. The text says 10K training data are collected from HaluEval-QA and TruthfulQA, categorized by thresholds, and then 'evaluate their performance on the HaluEval-QA dataset', without specifying a held-out split. If the switch model or the threshold γ is selected using test-set results, Figure 2 overstates the achievable trade-off. Please clarify the data split, and if thresholds are tuned, report validation-based selection or cross-validation.
  4. [§3.3, prompt template] The reward model prompt template includes the line 'Correct Answer: (Only provided when generating reward data.)'. This is reassuring, but the surrounding text in Section 3.3 says the ground truth answer is included as a reference 'for evaluation'. Please state explicitly that the reward model does not receive the ground truth when scoring MCTS rollouts at test time. If the correct answer is available to the reward model during search, the reported accuracy gains would be an artifact of test-set leakage.
minor comments (6)
  1. [§3.2] The acronym 'UTC' should be 'UCT'; it appears in the sentence introducing Equation (1).
  2. [Table 2] The table caption says 'on HaluEval-QA dataset' but the columns include TruthfulQA and SimpleQA; please correct the caption to describe the three English datasets.
  3. [References] Reference [44] is cited for ChineseFactEval, but the cited paper appears to be about Factool, a factuality detection framework; please verify that this is the correct source for the ChineseFactEval benchmark.
  4. [§4.1, Table 1] The tables report only point estimates. Given that several differences between MCTS and Best-of-N are small (e.g., ChineseSimpleQA with Qwen2: 36.00 vs 35.00), please report variance or significance tests over multiple runs to establish that the improvements are reliable.
  5. [Figure 2] Please add axis labels and units to Figure 2, and specify how 'average solving time per question' is measured (e.g., wall-clock time on which hardware).
  6. [Section 2.1 heading] The heading 'Halluciantion Mitigation' contains a typo; it should be 'Hallucination Mitigation'.

Circularity Check

0 steps flagged · score 0.0 of 10

No pith-circularity found: the MCTS reward-guided search is not defined in terms of the reported accuracy, and the GPT-4 dual role is a benchmark-validity confound rather than a reduction by construction.

full rationale

The paper's derivation chain is self-contained. HaluSearch defines node values via standard MCTS selection, expansion, evaluation, and backpropagation (Eqs. 1-4), with rewards supplied by a reward model; the final response is the highest-value path selected by greedy strategy. This is a search objective, not a quantity defined in terms of the reported accuracy. The reported accuracy is computed by GPT-4 comparing each generated response with the corresponding ground truth (Section 4.1). Although GPT-4 also serves as the reward model in the main experiments, the reward prompt scores hallucination likelihood without the correct answer, while the evaluation prompt verifies against the ground-truth answer; the reported quantity is therefore not the same variable as the optimized reward. The dual use of GPT-4 is a legitimate external-validity concern regarding possible judge alignment, but it is not a definitional equivalence or a fitted-parameter-renamed-as-prediction. The trained reward models in Section 4.3 are trained on a held-out subset (1,000 HaluEval-QA plus 500 TruthfulQA examples) and evaluated on the remaining data, so there is no direct label leakage into the test set. Self-citations, including HaluEval [39] and related technical reports [31,34], are used as benchmarks or related-work context rather than as a uniqueness theorem or as the sole justification of the method. The mismatch between the 'MCTS' row in Table 1 and the full HaluSearch/MCTSwitch system evaluated only in Section 4.4 is an overclaim about what was compared, but it is not a circular derivation. No step in the paper reduces by construction to its own input, so the circularity score is 0.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The central result depends on the model already knowing the facts (axiom 1), the sentence-level reward being a faithful guide (axiom 2), and GPT-4 reward scores tracking truth (axiom 3). Several algorithm hyperparameters are chosen by hand, and the switch threshold is tuned on the evaluation set.

free parameters (5)
  • UCT exploration weight w = 0.4
    Set by hand to balance exploration and exploitation; no sensitivity analysis is reported for this parameter.
  • Expansions per step K = 10
    Number of child nodes generated per selected node; varied in hyperparameter analysis only.
  • Rollouts per node m = 5
    Number of completed responses sampled to estimate node value.
  • Maximum simulations M = 20
    Total MCTS iterations; termination condition also includes reward threshold rth, which is unspecified.
  • Switch threshold gamma = 3, 4, 5 in Section 4.4; 0-5 in Figure 2
    Threshold on node value used to label switch training data and to control the slow-thinking ratio; selected on HaluEval-QA without a holdout.
assumptions (3)
  • domain assumption The policy LLM possesses the underlying factual knowledge for the test questions; hallucinations arise from ineffective generation rather than missing knowledge.
    Invoked in Section 3, citing [19, 33]; if the model cannot generate correct content through any path, tree search cannot recover it.
  • domain assumption Each sentence is a meaningful reasoning step, and the reward of completed rollouts propagated back to sentence nodes is a faithful estimate of future hallucination risk.
    Core modeling choice in Section 3.2; sentence-level decomposition and scalar rewards are assumed sufficient to steer the search.
  • domain assumption GPT-4 reward scores correlate with factual correctness and can be approximated by a trained reward model.
    Used throughout Section 3.3 and Section 4.1; if GPT-4 rewards are not aligned with truth, both search and final evaluation are biased.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Think More, Hallucinate Less: Mitigating Hallucinations via Dual Process of Fast and Slow Thinking." pith.science (2026). https://pith.science/paper/MHAKHLD3

@misc{pith2026250101306,
  author       = {Pith},
  title        = {Pith review of: Think More, Hallucinate Less: Mitigating Hallucinations via Dual Process of Fast and Slow Thinking},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MHAKHLD3}},
  note         = {Machine review of arXiv:2501.01306}
}
read the original abstract

Large language models (LLMs) demonstrate exceptional capabilities, yet still face the hallucination issue. Typical text generation approaches adopt an auto-regressive generation without deliberate reasoning, which often results in untrustworthy and factually inaccurate responses. In this paper, we propose HaluSearch, a novel framework that incorporates tree search-based algorithms (e.g. MCTS) to enable an explicit slow thinking generation process for mitigating hallucinations of LLMs during inference. Specifically, HaluSearch frames text generation as a step-by-step reasoning process, using a self-evaluation reward model to score each generation step and guide the tree search towards the most reliable generation pathway for fully exploiting the internal knowledge of LLMs. To balance efficiency and quality, we introduce a hierarchical thinking system switch mechanism inspired by the dual process theory in cognitive science, which dynamically alternates between fast and slow thinking modes at both the instance and step levels, adapting to the complexity of questions and reasoning states. We conduct extensive experiments on both English and Chinese datasets and the results show that our approach significantly outperforms baseline approaches.

Figures

Figures reproduced from arXiv: 2501.01306 by the authors.

Figure 1
Figure 1. The overview of our proposed HaluSearch approach. The left part demonstrates the process [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Impact of switch thresholds on time saving and accuracy on HaluEval-QA dataset. [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Results on HaluEval-QA w.r.t. the number of expanded nodes or the number of simulations. [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: An example showing the deliberative reasoning process with system switching of HaluSe [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 8 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Unveiling Knowledge Utilization Mechanisms in LLM-based Retrieval-Augmented Generation

    cs.CL 2025-05 conditional novelty 6.0 of 10

    RAG knowledge use in LLMs proceeds through four stages (refinement, elicitation, expression, contestation), and a new entropy metric (KAPE) identifies neurons whose deactivation shifts reliance between internal and ex...

  2. Reasoning on a Budget: A Survey of Adaptive and Controllable Test-Time Compute in LLMs

    cs.AI 2025-07 conditional novelty 5.0 of 10

    The survey's L1/L2 taxonomy and benchmark show that current reasoning models waste compute on easy problems and underthink hard ones, motivating more adaptive inference.

  3. PATS: Process-Level Adaptive Thinking Mode Switching

    cs.CL 2025-05 conditional novelty 5.0 of 10

    PATS adapts the number of beam-search candidates per reasoning step using process reward model scores, improving accuracy-efficiency tradeoffs on math benchmarks.

  4. Incentivizing Dual Process Thinking for Efficient Large Language Model Reasoning

    cs.AI 2025-05 conditional novelty 5.0 of 10

    ACPO uses special fast/slow thinking tokens and an online length-budget reward to make LLM math reasoning shorter and difficulty-adaptive, with modest accuracy loss.

  5. Reason4Rec: Deliberative User Preference Alignment of Large Language Models for Recommendation

    cs.IR 2025-02 conditional novelty 5.0 of 10

    An LLM recommender that decomposes preference reasoning into summarization, matching, and prediction steps, trained on review text, improves rating accuracy and explanation quality on Amazon and Yelp benchmarks.

  6. A Comprehensive Survey on Trustworthiness in Reasoning with Large Language Models

    cs.CL 2025-09 conditional novelty 4.0 of 10

    A structured literature survey concluding that reasoning capabilities do not automatically make LLMs more trustworthy and can introduce new vulnerabilities in safety, robustness, and privacy.

  7. Towards Concise and Adaptive Thinking in Large Reasoning Models: A Survey

    cs.AI 2025-07 conditional novelty 3.0 of 10

    A comprehensive review that categorizes methods for shortening and adaptively triggering chain-of-thought reasoning in large language models.

  8. Generative AI Act II: Test Time Scaling Drives Cognition Engineering

    cs.CL 2025-04 conditional novelty 3.0 of 10

    Test-time scaling techniques such as long chain-of-thought, tree search, and self-correction define the paper's 'cognition engineering' paradigm, which it surveys, taxonomizes, and tutorials.

Reference graph

Works this paper leans on

46 extracted references · 12 canonical work pages · cited by 8 Pith papers

  1. [1]

    A survey of large language models

    Wayne Xin Zhao, Kun Zhou, Junyi Li, Tianyi Tang, Xiaolei Wang, Yupeng Hou, Yingqian Min, Beichen Zhang, Junjie Zhang, Zican Dong, Yifan Du, Chen Yang, Yushuo Chen, Zhipeng Chen, Jinhao Jiang, Ruiyang Ren, Yifan Li, Xinyu Tang, Zikang Liu, Peiyu Liu, Jian-Yun Nie, and Ji-Rong Wen. A survey of large language models. CoRR, abs/2303.18223, 2023

    arXiv 2023

  2. [2]

    Survey of hallucination in natural language generation

    Ziwei Ji, Nayeon Lee, Rita Frieske, Tiezheng Yu, Dan Su, Yan Xu, Etsuko Ishii, Ye Jin Bang, Andrea Madotto, and Pascale Fung. Survey of hallucination in natural language generation. ACM Computing Surveys, 55(12):1–38, 2023

    2023

  3. [3]

    A survey on hallucination in large language models: Principles, taxonomy, challenges, and open questions

    Lei Huang, Weijiang Yu, Weitao Ma, Weihong Zhong, Zhangyin Feng, Haotian Wang, Qiang- long Chen, Weihua Peng, Xiaocheng Feng, Bing Qin, and Ting Liu. A survey on hallucination in large language models: Principles, taxonomy, challenges, and open questions. CoRR, abs/2311.05232, 2023

    arXiv 2023

  4. [4]

    Sheth, and Amitava Das

    Vipula Rawte, Amit P. Sheth, and Amitava Das. A survey of hallucination in large foundation models. CoRR, abs/2309.05922, 2023

    arXiv 2023

  5. [5]

    Cognitive mirage: A review of hallucinations in large language models

    Hongbin Ye, Tong Liu, Aijia Zhang, Wei Hua, and Weiqiang Jia. Cognitive mirage: A review of hallucinations in large language models. CoRR, abs/2309.06794, 2023

    arXiv 2023

  6. [6]

    Siren’s song in the AI ocean: A survey on hallucination in large language models

    Yue Zhang, Yafu Li, Leyang Cui, Deng Cai, Lemao Liu, Tingchen Fu, Xinting Huang, Enbo Zhao, Yu Zhang, Yulong Chen, Longyue Wang, Anh Tuan Luu, Wei Bi, Freda Shi, and Shuming Shi. Siren’s song in the AI ocean: A survey on hallucination in large language models. CoRR, abs/2309.01219, 2023

    arXiv 2023

  7. [7]

    Kankanhalli

    Ziwei Xu, Sanjay Jain, and Mohan S. Kankanhalli. Hallucination is inevitable: An innate limitation of large language models. CoRR, abs/2401.11817, 2024

    arXiv 2024

  8. [8]

    Llms will always hallucinate, and we need to live with this

    Sourav Banerjee, Ayushi Agarwal, and Saloni Singla. Llms will always hallucinate, and we need to live with this. CoRR, abs/2409.05746, 2024

    arXiv 2024

Show all 46 references
  1. [9]

    Textbooks are all you need II: phi-1.5 technical report

    Yuanzhi Li, Sébastien Bubeck, Ronen Eldan, Allie Del Giorno, Suriya Gunasekar, and Yin Tat Lee. Textbooks are all you need II: phi-1.5 technical report. CoRR, abs/2309.05463, 2023

    2023 arXiv

  2. [10]

    Manning, and Chelsea Finn

    Katherine Tian, Eric Mitchell, Huaxiu Yao, Christopher D. Manning, and Chelsea Finn. Fine- tuning language models for factuality. In The Twelfth International Conference on Learning Representations, ICLR 2024, Vienna, Austria, May 7-11, 2024 . OpenReview.net, 2024

    2024

  3. [11]

    Halo: Estimation and reduction of hallucinations in open-source weak large language models

    Mohamed Elaraby, Mengyin Lu, Jacob Dunn, Xueying Zhang, Yu Wang, and Shizhu Liu. Halo: Estimation and reduction of hallucinations in open-source weak large language models. CoRR, abs/2308.11764, 2023

    2023 arXiv

  4. [12]

    FLAME: factuality-aware alignment for large language models

    Sheng-Chieh Lin, Luyu Gao, Barlas Oguz, Wenhan Xiong, Jimmy Lin, Wen-tau Yih, and Xilun Chen. FLAME: factuality-aware alignment for large language models. CoRR, abs/2405.01525, 2024

    2024 arXiv

  5. [13]

    Chain-of-verification reduces hallucination in large language models

    Shehzaad Dhuliawala, Mojtaba Komeili, Jing Xu, Roberta Raileanu, Xian Li, Asli Celikyilmaz, and Jason Weston. Chain-of-verification reduces hallucination in large language models. In Lun-Wei Ku, Andre Martins, and Vivek Srikumar, editors, Findings of the Association for Comput...

    2024

  6. [14]

    Self-refine: Iterative refinement with self-feedback

    Aman Madaan, Niket Tandon, Prakhar Gupta, Skyler Hallinan, Luyu Gao, Sarah Wiegreffe, Uri Alon, Nouha Dziri, Shrimai Prabhumoye, Yiming Yang, et al. Self-refine: Iterative refinement with self-feedback. Advances in Neural Information Processing Systems , 36, 2024

    2024

  7. [15]

    Ever: Mitigating hallucination in large language models through real-time verification and rectification

    Haoqiang Kang, Juntong Ni, and Huaxiu Yao. Ever: Mitigating hallucination in large language models through real-time verification and rectification. CoRR, abs/2311.09114, 2023. 11

    2023 arXiv

  8. [16]

    Retrieval-augmented generation for knowledge-intensive nlp tasks

    Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, et al. Retrieval-augmented generation for knowledge-intensive nlp tasks. Advances in Neural Information Processing Systems...

    2020

  9. [17]

    Self-rag: Learn- ing to retrieve, generate, and critique through self-reflection

    Akari Asai, Zeqiu Wu, Yizhong Wang, Avirup Sil, and Hannaneh Hajishirzi. Self-rag: Learn- ing to retrieve, generate, and critique through self-reflection. In The Twelfth International Conference on Learning Representations, ICLR 2024, Vienna, Austria, May 7-11, 2024 . Open- Re...

    2024

  10. [18]

    Chain-of-thought prompting elicits reasoning in large language models

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, et al. Chain-of-thought prompting elicits reasoning in large language models. Advances in neural information processing systems , 35:24824–24837, 2022

    2022

  11. [19]

    Le, Ed H

    Xuezhi Wang, Jason Wei, Dale Schuurmans, Quoc V . Le, Ed H. Chi, Sharan Narang, Aakanksha Chowdhery, and Denny Zhou. Self-consistency improves chain of thought reasoning in language models. In The Eleventh International Conference on Learning Representations, ICLR 2023, Kigali...

    2023

  12. [20]

    Thinking, fast and slow

    Daniel Kahneman. Thinking, fast and slow. Farrar , Straus and Giroux, 2011

    2011

  13. [21]

    Dual processes in reasoning? Cognition, 3(2):141–154, 1974

    Peter C Wason and J St BT Evans. Dual processes in reasoning? Cognition, 3(2):141–154, 1974

    1974

  14. [22]

    The dawn after the dark: An empirical study on factuality hallucination in large language models

    Junyi Li, Jie Chen, Ruiyang Ren, Xiaoxue Cheng, Xin Zhao, Jian-Yun Nie, and Ji-Rong Wen. The dawn after the dark: An empirical study on factuality hallucination in large language models. In Lun-Wei Ku, Andre Martins, and Vivek Srikumar, editors,Proceedings of the 62nd Annual M...

    2024

  15. [23]

    Fung, Jing Li, Manling Li, and Heng Ji

    Yuji Zhang, Sha Li, Jiateng Liu, Pengfei Yu, Yi R. Fung, Jing Li, Manling Li, and Heng Ji. Knowledge overshadowing causes amalgamated hallucination in large language models. CoRR, abs/2407.08039, 2024

    2024 arXiv

  16. [24]

    A pretrainer’s guide to training data: Measuring the effects of data age, domain coverage, quality, & toxicity

    Shayne Longpre, Gregory Yauney, Emily Reif, Katherine Lee, Adam Roberts, Barret Zoph, Denny Zhou, Jason Wei, Kevin Robinson, David Mimno, and Daphne Ippolito. A pretrainer’s guide to training data: Measuring the effects of data age, domain coverage, quality, & toxicity. In Kev...

    2024

  17. [25]

    Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll L. Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, John Schulman, Jacob Hilton, Fraser Kelton, Luke Miller, Maddie Simens, Amanda Askell, Peter Welinder, Paul F. Christiano, Jan Leik...

    2022

  18. [26]

    Let’s verify step by step

    Hunter Lightman, Vineet Kosaraju, Yuri Burda, Harrison Edwards, Bowen Baker, Teddy Lee, Jan Leike, John Schulman, Ilya Sutskever, and Karl Cobbe. Let’s verify step by step. In The Twelfth International Conference on Learning Representations, ICLR 2024, Vienna, Austria, May 7-1...

    2024

  19. [27]

    Tree of thoughts: Deliberate problem solving with large language models

    Shunyu Yao, Dian Yu, Jeffrey Zhao, Izhak Shafran, Tom Griffiths, Yuan Cao, and Karthik Narasimhan. Tree of thoughts: Deliberate problem solving with large language models. Ad- vances in Neural Information Processing Systems , 36, 2024

    2024

  20. [28]

    Scaling LLM test-time compute optimally can be more effective than scaling model parameters

    Charlie Snell, Jaehoon Lee, Kelvin Xu, and Aviral Kumar. Scaling LLM test-time compute optimally can be more effective than scaling model parameters. CoRR, abs/2408.03314, 2024

    2024 arXiv

  21. [29]

    Q*: Improving multi-step reasoning for llms with deliberative planning

    Chaojie Wang, Yanchen Deng, Zhiyi Lv, Zeng Liang, Jujie He, Shuicheng Yan, and Bo An. Q*: Improving multi-step reasoning for llms with deliberative planning. CoRR, abs/2406.14283, 2024

    2024 arXiv

  22. [30]

    Mindstar: Enhancing math reasoning in pre-trained llms at inference time

    Jikun Kang, Xin Zhe Li, Xi Chen, Amirreza Kazemi, and Boxing Chen. Mindstar: Enhancing math reasoning in pre-trained llms at inference time. CoRR, abs/2405.16265, 2024. 12

    2024 arXiv

  23. [31]

    Imitate, explore, and self-improve: A reproduction report on slow-thinking reasoning systems

    Yingqian Min, Zhipeng Chen, Jinhao Jiang, Jie Chen, Jia Deng, Yiwen Hu, Yiru Tang, Jiapeng Wang, Xiaoxue Cheng, Huatong Song, et al. Imitate, explore, and self-improve: A reproduction report on slow-thinking reasoning systems. arXiv preprint arXiv:2412.09413, 2024

    2024 arXiv

  24. [32]

    Inference- time intervention: Eliciting truthful answers from a language model

    Kenneth Li, Oam Patel, Fernanda Viégas, Hanspeter Pfister, and Martin Wattenberg. Inference- time intervention: Eliciting truthful answers from a language model. Advances in Neural Information Processing Systems, 36, 2024

    2024

  25. [33]

    Llms know more than they show: On the intrinsic representation of LLM hallucinations

    Hadas Orgad, Michael Toker, Zorik Gekhman, Roi Reichart, Idan Szpektor, Hadas Kotek, and Yonatan Belinkov. Llms know more than they show: On the intrinsic representation of LLM hallucinations. CoRR, abs/2410.02707, 2024

    2024 arXiv

  26. [34]

    Technical report: Enhancing llm reasoning with reward-guided tree search

    Jinhao Jiang, Zhipeng Chen, Yingqian Min, Jie Chen, Xiaoxue Cheng, Jiapeng Wang, Yiru Tang, Haoxiang Sun, Jia Deng, Wayne Xin Zhao, et al. Technical report: Enhancing llm reasoning with reward-guided tree search. arXiv preprint arXiv:2411.11694, 2024

    2024 arXiv

  27. [35]

    Bandit based monte-carlo planning

    Levente Kocsis and Csaba Szepesvári. Bandit based monte-carlo planning. In Johannes Fürnkranz, Tobias Scheffer, and Myra Spiliopoulou, editors, Machine Learning: ECML 2006, 17th European Conference on Machine Learning, Berlin, Germany, September 18-22, 2006, Proceedings, volum...

    2006

  28. [36]

    Rosé, Daniel Fried, and Hejia Zhang

    Yiqing Xie, Wenxuan Zhou, Pradyot Prakash, Di Jin, Yuning Mao, Quintin Fettes, Arya Tale- bzadeh, Sinong Wang, Han Fang, Carolyn P. Rosé, Daniel Fried, and Hejia Zhang. Improving model factuality with fine-grained critique-based evaluator. CoRR, abs/2410.18359, 2024

    arXiv

  29. [37]

    A technique for the measurement of attitudes

    Rensis Likert. A technique for the measurement of attitudes. Archives of Psychology, 1932

    1932

  30. [38]

    Distilling system 2 into system 1

    Ping Yu, Jing Xu, Jason Weston, and Ilia Kulikov. Distilling system 2 into system 1. CoRR, abs/2407.06023, 2024

    2024 arXiv

  31. [39]

    Halueval: A large-scale hallucination evaluation benchmark for large language models

    Junyi Li, Xiaoxue Cheng, Wayne Xin Zhao, Jian-Yun Nie, and Ji-Rong Wen. Halueval: A large-scale hallucination evaluation benchmark for large language models. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing , pages 6449–6464, 2023

    2023

  32. [40]

    Truthfulqa: Measuring how models mimic human falsehoods

    Stephanie Lin, Jacob Hilton, and Owain Evans. Truthfulqa: Measuring how models mimic human falsehoods. In Smaranda Muresan, Preslav Nakov, and Aline Villavicencio, editors, Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (V olume 1: Long...

    2022

  33. [41]

    Measuring short-form factuality in large language models

    Jason Wei, Nguyen Karina, Hyung Won Chung, Yunxin Joy Jiao, Spencer Papay, Amelia Glaese, John Schulman, and William Fedus. Measuring short-form factuality in large language models. arXiv preprint arXiv:2411.04368, 2024

    2024 arXiv

  34. [42]

    Evaluating hallucinations in chinese large language models

    Qinyuan Cheng, Tianxiang Sun, Wenwei Zhang, Siyin Wang, Xiangyang Liu, Mozhi Zhang, Jun- liang He, Mianqiu Huang, Zhangyue Yin, Kai Chen, and Xipeng Qiu. Evaluating hallucinations in chinese large language models. CoRR, abs/2310.03368, 2023

    arXiv

  35. [43]

    Chinese simpleqa: A chinese factuality evaluation for large language models

    Yancheng He, Shilong Li, Jiaheng Liu, Yingshui Tan, Weixun Wang, Hui Huang, Xingyuan Bu, Hangyu Guo, Chengwei Hu, Boren Zheng, Zhuoran Lin, Xuepeng Liu, Dekai Sun, Shirong Lin, Zhicheng Zheng, Xiaoyong Zhu, Wenbo Su, and Bo Zheng. Chinese simpleqa: A chinese factuality evaluat...

    2024 arXiv

  36. [44]

    Factool: Factuality detection in generative AI - A tool augmented framework for multi-task and multi-domain scenarios

    I-Chun Chern, Steffi Chern, Shiqi Chen, Weizhe Yuan, Kehua Feng, Chunting Zhou, Junxian He, Graham Neubig, and Pengfei Liu. Factool: Factuality detection in generative AI - A tool augmented framework for multi-task and multi-domain scenarios. CoRR, abs/2307.13528, 2023

    2023 arXiv

  37. [45]

    Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, Anirudh Goyal, Anthony Hartshorn, Aobo Yang, Archi Mitra, Archie Sravankumar, Artem Korenev, Arthur Hinsvark, Arun Rao, Aston Zha...

    2024 arXiv

  38. [46]

    Qwen2 technical report

    An Yang, Baosong Yang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Zhou, Chengpeng Li, Chengyuan Li, Dayiheng Liu, Fei Huang, Guanting Dong, Haoran Wei, Huan Lin, Jialong Tang, Jialin Wang, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Ma, Jianxin Yang, Jin Xu, Jingren Zhou, Jinze...

    2024 arXiv

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.