Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

A Survey on Training-free Alignment of Large Language Models

T0 review · 3 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read Training-free alignment matches fine-tuned results, survey argues

desk verdict Useful survey with a sensible taxonomy, but the headline 'TF beats FT' claim is unsupported because Table 1's FT baseline is itself training-free. read the letter →

arxiv 2508.09016 v4 pith:S4F4ODX3 submitted 2025-08-12 cs.CL cs.LG

Birong Pan , Yongqi Li , Weiyu Zhang , Wenpeng Lu , Mayi Xu , Shen Zhou , Yuanyuan Zhu , Ming Zhong
show 1 more author
Tieyun Qian
This is my paper
classification cs.CLcs.LG
keywords training-freealignmentlargelanguagemodelsin-contextlearningdecoding-timesafedecodingLLMsafetymultimodalsurvey
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

This survey argues that large language models can be aligned with human values without fine-tuning, by changing the prompt, steering the decoding process, or filtering the final output. It organizes the rapidly growing field of training-free alignment methods into three stages - pre-decoding, in-decoding, and post-decoding - and claims that such methods can match or even exceed the safety and helpfulness of fine-tuned alignment. If true, this matters because it would allow users of closed-source or resource-constrained models to align them cheaply, without retraining or parameter access. The paper also catalogs the main weaknesses of each stage, such as generalization limits and added latency, and proposes an agenda for making training-free alignment more robust and controllable.

What carries the argument

The organizing device is a three-stage taxonomy of the generation pipeline: pre-decoding interventions (prompt engineering, in-context examples, input detectors), in-decoding adjustments (modifying hidden states, computing logits differences, reward-guided search), and post-decoding refinements (self-examination, filtering, and correction of complete outputs). This taxonomy is the paper's framework for comparing methods, and the quantitative comparison in Table 1 grounds the claim that training-free alignment can match fine-tuning.

What would settle it

Run the same three training-free methods (URIAL, SCANS, RA-LLM) against SafeDecoding across several additional open-weight models such as Llama-3-8B and Mistral-7B, evaluating with human annotators and multiple safety classifiers; if fine-tuning wins on most tasks and models, the paper's central claim would fail.

Watch

Extended reading notes

Core claim

The paper's central claim is that training-free alignment - methods that modify prompts, steer decoding, or filter outputs rather than updating weights - can match or even exceed the safety and helpfulness performance of fine-tuning. The evidence is a single quantitative comparison on llama2-7b-chat: SCANS, a decoding-time activation-steering method, outperforms the fine-tuned SafeDecoding on SafeEdit and TruthfulQA and matches it on AdvBench, while the post-decoding method RA-LLM matches or exceeds the fine-tuned baseline on safety metrics. The authors read this as evidence that training-free alignment is a viable supplement to fine-tuning, especially in closed-source, low-resource, or know

Load-bearing premise

The central comparison rests on one model (llama2-7b-chat), three benchmarks, and one safety classifier, and the paper generalizes from that single table to the broad claim that training-free alignment can match or exceed fine-tuning.

Editorial extensions

If this is right

  • Users of closed-source or proprietary models can align them through prompts, decoding steering, and output filtering - no parameter access needed.
  • Because training-free methods avoid weight updates, they tend to preserve pretrained knowledge better than fine-tuning, which the paper shows through lower benign refusal rates on TruthfulQA.
  • The taxonomy implies alignment can be treated as a modular intervention at any pipeline stage, letting practitioners trade off safety, helpfulness, latency, and model-access requirements.
  • Pre-decoding and post-decoding methods work in black-box settings but face generalization limits and latency; in-decoding methods are stronger but require internal access - so no single training-free method fits all scenarios.

Reading between the lines

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

  • One consequence the authors leave implicit is that if training-free alignment genuinely matches fine-tuning on safety and helpfulness, the expensive fine-tuning step in alignment pipelines could shrink to a knowledge-plus-capability stage, with safety handled at inference time.
  • The three-stage taxonomy suggests that future work might treat alignment as a search problem over the intervention space - which stage, which steering strength, which filter - rather than as a fixed training recipe.
  • A testable extension would be to compose methods across stages, since the survey evaluates each stage in isolation; prompt-level defense plus decoding-time steering plus post-hoc filtering might outperform any single method.
  • The paper's reliance on a single safety classifier suggests results could shift if the evaluation metric changes, so a robustness check with multiple classifiers and human judgments would be a direct next step.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper surveys training-free (TF) alignment methods for large language models and multimodal large language models. It proposes a taxonomy of pre-decoding, in-decoding, and post-decoding interventions, reviews representative methods in each stage, discusses their mechanisms and limitations, and reports a small comparative experiment on llama2-7b-chat that is intended to show that TF alignment can match or exceed fine-tuning (FT) alignment in safety and helpfulness. The paper concludes with open challenges and future research directions, including general-capability preservation, inference overhead, generalization, controllability, and multimodal output alignment.

Significance. The survey addresses a timely and rapidly growing area and provides a useful organizing framework. The pre-/in-/post-decoding taxonomy is intuitive, and the coverage spans both unimodal LLMs and MLLMs, with a broad set of cited methods. The appendix comparison of methods along accessibility, storage, efficiency, and generalization dimensions (Table 2) is practical and valuable. If the quantitative claim were properly supported, it would be important for practitioners choosing alignment strategies under resource or access constraints. However, the current experimental support for the headline claim is not valid, and the survey's descriptive content is stronger than its comparative conclusion.

major comments (3)
  1. [§3.4, Table 1; abstract; §2.2; §A.2] The 'FT Alignment' comparator SafeDecoding (Xu et al., 2024b) is not a fine-tuning method. Its published mechanism is safety-aware decoding: at inference time it uses an expert model to identify safety-critical token sets and adjusts token probabilities, without updating any model parameters. Labeling it 'FT Alignment' in Table 1 and using it as the sole fine-tuning representative invalidates the conclusion that TF methods 'can match or even exceed' FT methods. The table actually compares four TF/decoding-time methods. In addition, the 'Defaults' row is llama2-7b-chat, which is itself an RLHF-fine-tuned model, so no row represents a genuinely fine-tuned baseline versus an unfine-tuned one. The same unsupported inference is repeated in §2.2 and §A.2 ('FT alignment methods cause the most severe knowledge impairment'). This is load-bearing: the paper's headline comparative claim rests on th
  2. [§3.4 and Limitations] Even setting the misclassification aside, the quantitative evidence is a single model (llama2-7b-chat), three benchmarks, no error bars or significance tests, and a safety classifier borrowed from the SafeEdit paper (Wang et al., 2024a). The Limitations acknowledge the single-model scope, but the abstract and §2.2 state the match/exceed conclusion without that caveat. Please either present Table 1 explicitly as an illustrative case study or add the necessary scope conditions and uncertainty quantification before making a general comparative claim.
  3. [§3.3, Figure 1, §2.2] The survey's use of 'training-free' is broader than the phrase 'no training overheads' in §2.2. For instance, Aligner (Ji et al., 2024a) trains a separate correction model, and CA VGAN (Li et al., 2025c) trains a GAN on internal representations. These are listed as TF alignment because the target LLM is not fine-tuned. This definition is defensible, but the paper should state it explicitly; otherwise the abstract's 'without heavily retraining LLMs' and the §2.2 claim of 'no training overheads' are misleading. This is not fatal, but it affects the taxonomy's clarity.
minor comments (5)
  1. [§3.4] The sentence 'For FT alignment methods, we select SafeDecoding...' is contradicted by the cited paper's own description; this should be corrected regardless of the experimental outcome.
  2. [§A.1] The evaluation description says 'If responses to safety questions ... contain more refusal-related keywords', but TruthfulQA is described as 753 benign questions. Please adjust the wording to avoid implying that TruthfulQA contains safety questions.
  3. [§4.2] The subsection title 'TF Alignment for Uni-Modal Model' appears to be a typo; the content discusses extending alignment to models with multimodal output. Consider renaming to clarify the intended scope.
  4. [Table 2] The legend uses the symbol ' and %' which likely lost the checkmark/cross glyphs in typesetting. Please ensure the table symbols are rendered consistently and explained.
  5. [Ethics Statement] The statement that 'there will not be any negative social impacts' is too absolute for a paper discussing safety methods; a more measured phrasing would be appropriate.

Circularity Check

1 steps flagged · score 6.0 of 10

Central TF-vs-FT claim rests on a table whose only 'FT' row is itself a training-free decoding-time method, making the comparison TF-vs-TF by the paper's own definitions.

  1. self definitional [Abstract; §3.4 'Quantitative Analysis'; Table 1 (repeated in §A.2)]
    "Abstract: 'training-free (TF) alignment techniques--leveraging in-context learning, decoding-time adjustments, and post-generation corrections--offer a promising alternative...' §3.4: 'For FT alignment methods, we select SafeDecoding (Xu et al., 2024b), a relatively new method that accounts for jailbreak attacks...' Table 1: 'SafeDecoding FT Alignment 100.00 94.60 54.44'."

    The abstract defines TF as including decoding-time adjustments, and §3.2 defines in-decoding TF as 'adjusting token selection during generation.' SafeDecoding is a safety-aware decoding strategy that alters token probabilities/generation at inference time without updating model parameters, so by the paper's own taxonomy it belongs to the in-decoding TF class. Yet §3.4 installs it as the sole 'FT Alignment' row and then concludes that 'TF alignment methods can match or even exceed the safety and helpfulness performance of FT alignment methods.' With the only fine-tuning comparator being itself a TF method, the headline comparison reduces to TF-vs-TF; the claimed TF≥FT result is an artifact of the label rather than a measured comparison against fine-tuning. The conclusion may be true, but th

full rationale

This is a survey rather than a derivation, so most of its content (taxonomy, method summaries, future directions) is not circular. Citations to URIAL and other external works provide independent support for the general possibility that prompting/decoding methods can rival tuned models; the survey's own organization does not fit parameters and then predict them. However, the paper's strongest and most repeated claim—TF alignment can match or exceed FT alignment—is supported in §3.4/Table 1 by a comparison in which the only 'FT Alignment' method, SafeDecoding, is by the paper's own definition a training-free, decoding-time adjustment method. Thus the central empirical claim as presented is partially circular/constructed: the FT baseline is a TF method in disguise. The Limitations section's caveats (single model llama2-7b-chat, one evaluation protocol, base models needing prior fine-tuning) compound the problem but are secondary to the mislabeling. Because the claim has external support (URIAL, cited in §2.2), the circularity is partial rather than total, giving a score of 6.

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

The paper introduces no new mathematical axioms or entities. Its load-bearing assumptions are definitional choices about what counts as 'training-free' and the representativeness of the evaluation setup. The authors acknowledge the base-model requirement in the Limitations, which partially mitigates the overreach, but the assumptions remain structural to the paper's claims.

assumptions (3)
  • domain assumption The taxonomy of pre-decoding, in-decoding, and post-decoding is a valid and complete partition of training-free alignment methods.
    The paper asserts this partition without formal justification or a coverage analysis. Some methods (e.g., PICA) use both prompt engineering and hidden states, so the categories are not strictly disjoint.
  • ad hoc to paper Methods that train a separate proxy model (e.g., Aligner) are still considered 'training-free' as long as the target LLM is not fine-tuned.
    This definitional choice broadens the scope of the survey and aligns with the title, but it is not defended. It appears implicitly in Section 3.3 when Aligner is included despite training a separate correction model.
  • domain assumption The safety classifier from Wang et al. (2024a) and the benchmark choices (AdvBench, SafeEdit, TruthfulQA) provide a sufficient measure of alignment quality.
    The evaluation in Section 3.4 relies entirely on these external tools without validation or discussion of their limitations. The helpfulness measure (Benign Refusing Rate) is a simple keyword heuristic that may not capture actual utility.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Survey on Training-free Alignment of Large Language Models." pith.science (2026). https://pith.science/paper/S4F4ODX3

@misc{pith2026250809016,
  author       = {Pith},
  title        = {Pith review of: A Survey on Training-free Alignment of Large Language Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/S4F4ODX3}},
  note         = {Machine review of arXiv:2508.09016}
}
read the original abstract

The alignment of large language models (LLMs) aims to ensure their outputs adhere to human values, ethical standards, and legal norms. Traditional alignment methods often rely on resource-intensive fine-tuning (FT), which may suffer from knowledge degradation and face challenges in scenarios where the model accessibility or computational resources are constrained. In contrast, training-free (TF) alignment techniques--leveraging in-context learning, decoding-time adjustments, and post-generation corrections--offer a promising alternative by enabling alignment without heavily retraining LLMs, making them adaptable to both open-source and closed-source environments. This paper presents the first systematic review of TF alignment methods, categorizing them by stages of pre-decoding, in-decoding, and post-decoding. For each stage, we provide a detailed examination from the viewpoint of LLMs and multimodal LLMs (MLLMs), highlighting their mechanisms and limitations. Furthermore, we identify key challenges and future directions, paving the way for more inclusive and effective TF alignment techniques. By synthesizing and organizing the rapidly growing body of research, this survey offers a guidance for practitioners and advances the development of safer and more reliable LLMs.

Figures

Figures reproduced from arXiv: 2508.09016 by the authors.

Figure 1
Figure 1. Taxonomy of training-free (TF) alignment methodologies for LLMs, categorized into pre-decoding, [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. A conceptual framework illustrating training-free (TF) alignment strategies for large language models [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

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

  1. SafeGene: Reusable Adapters for Transferable Safety Alignment

    cs.AI 2026-06 unverdicted novelty 5.0 of 10

    SafeGene extracts task-transferable safety vectors from model discrepancies and applies them through layer-wise recalibration to reduce harmful outputs in downstream-adapted LLMs without retraining.

Reference graph

Works this paper leans on

112 extracted references · 18 canonical work pages · cited by 1 Pith paper

  1. [1]

    online" 'onlinestring :=

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint eprinttype howpublished institution journal key month note number organization pages publisher school series title type volume year doi pubmed url lastchecked label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block STRING...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Marwa Abdulhai, Gregory Serapio-Garc \'i a, Clement Crepy, Daria Valter, John Canny, and Natasha Jaques. 2024. https://doi.org/10.18653/v1/2024.emnlp-main.982 Moral foundations of large language models . In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 17737--17752, Miami, Florida, USA. Association for Compu...

    doi:10.18653/v1/2024.emnlp-main.982 2024

  4. [4]

    Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, et al. 2023. Gpt-4 technical report. arXiv preprint arXiv:2303.08774

    arXiv 2023

  5. [5]

    Badr AlKhamissi, Muhammad ElNokrashy, Mai Alkhamissi, and Mona Diab. 2024. https://doi.org/10.18653/v1/2024.acl-long.671 Investigating cultural alignment of large language models . In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 12404--12422, Bangkok, Thailand. Association for Compu...

    doi:10.18653/v1/2024.acl-long.671 2024

  6. [6]

    Jinze Bai, Shuai Bai, Shusheng Yang, Shijie Wang, Sinan Tan, Peng Wang, Junyang Lin, Chang Zhou, and Jingren Zhou. 2023. Qwen-vl: A frontier large vision-language model with versatile abilities. arXiv preprint arXiv:2308.12966

    arXiv 2023

  7. [7]

    Michael Beukman, Devon Jarvis, Richard Klein, Steven James, and Benjamin Rosman. 2023. https://openreview.net/forum?id=PJhjkSFlbG Dynamics generalisation in reinforcement learning via adaptive context-aware policies . In Thirty-seventh Conference on Neural Information Processing Systems

    2023

  8. [8]

    Amrita Bhattacharjee, Shaona Ghosh, Traian Rebedea, and Christopher Parisien. 2024. https://arxiv.org/abs/2410.01174 Towards inference-time category-wise safety steering for large language models . Preprint, arXiv:2410.01174

    arXiv 2024

Show all 112 references
  1. [9]

    Bochuan Cao, Yuanpu Cao, Lu Lin, and Jinghui Chen. 2024. https://doi.org/10.18653/v1/2024.acl-long.568 Defending against alignment-breaking attacks via robustly aligned LLM . In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: ...

    2024 doi

  2. [10]

    Zouying Cao, Yifei Yang, and Hai Zhao. 2025. Scans: Mitigating the exaggerated safety for llms via safety-conscious activation steering. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, pages 23523--23531

    2025

  3. [11]

    Lin Chen, Jinsong Li, Xiaoyi Dong, Pan Zhang, Conghui He, Jiaqi Wang, Feng Zhao, and Dahua Lin. 2025. Sharegpt4v: Improving large multi-modal models with better captions. In European Conference on Computer Vision, pages 370--387. Springer

    2025

  4. [12]

    Meiqi Chen, Yubo Ma, Kaitao Song, Yixin Cao, Yan Zhang, and Dongsheng Li. 2024 a . https://doi.org/10.18653/v1/2024.acl-long.512 Improving large language models in event relation logical prediction . In Proceedings of the 62nd Annual Meeting of the Association for Computationa...

    2024 doi

  5. [13]

    Ruizhe Chen, Xiaotian Zhang, Meng Luo, Wenhao Chai, and Zuozhu Liu. 2024 b . https://arxiv.org/abs/2410.04070 Pad: Personalized alignment of llms at decoding-time . Preprint, arXiv:2410.04070

    2024 arXiv

  6. [14]

    Yang Chen, Ethan Mendes, Sauvik Das, Wei Xu, and Alan Ritter. 2023. https://arxiv.org/abs/2310.02224 Can language models be instructed to protect personal information? Preprint, arXiv:2310.02224

    2023 arXiv

  7. [15]

    Jiale Cheng, Xiao Liu, Kehan Zheng, Pei Ke, Hongning Wang, Yuxiao Dong, Jie Tang, and Minlie Huang. 2024. https://doi.org/10.18653/v1/2024.acl-long.176 Black-box prompt optimization: Aligning large language models without model training . In Proceedings of the 62nd Annual Meet...

    2024 doi

  8. [16]

    Wei-Lin Chiang, Zhuohan Li, Zi Lin, Ying Sheng, Zhanghao Wu, Hao Zhang, Lianmin Zheng, Siyuan Zhuang, Yonghao Zhuang, Joseph E Gonzalez, et al. 2023. Vicuna: An open-source chatbot impressing gpt-4 with 90\ See https://vicuna. lmsys. org (accessed 14 April 2023), 2(3):6

    2023

  9. [17]

    Yi Ding, Bolian Li, and Ruqi Zhang. 2024. https://arxiv.org/abs/2410.06625 Eta: Evaluating then aligning safety of vision language models at inference time . Preprint, arXiv:2410.06625

    2024 arXiv

  10. [18]

    Xuefeng Du, Reshmi Ghosh, Robert Sim, Ahmed Salem, Vitor Carvalho, Emily Lawton, Yixuan Li, and Jack W. Stokes. 2024. https://arxiv.org/abs/2410.00296 Vlmguard: Defending vlms against malicious prompts via unlabeled data . Preprint, arXiv:2410.00296

    2024 arXiv

  11. [19]

    Shitong Duan, Xiaoyuan Yi, Peng Zhang, Tun Lu, Xing Xie, and Ning Gu. 2024. https://openreview.net/forum?id=m3RRWWFaVe DENEVIL : TOWARDS DECIPHERING AND NAVIGATING THE ETHICAL VALUES OF LARGE LANGUAGE MODELS VIA INSTRUCTION LEARNING . In The Twelfth International Conference on...

    2024

  12. [20]

    Hainan Fang, Di Huang, Yuanbo Wen, Yunpu Zhao, Tonghui He, QiCheng Wang, Shuo Wang, Rui Zhang, and Qi Guo. 2024. https://openreview.net/forum?id=cayKVPCrOP GOOD : Decoding-time black-box LLM alignment

    2024

  13. [21]

    Yu Fei, Yasaman Razeghi, and Sameer Singh. 2024. https://arxiv.org/abs/2410.09300 Nudging: Inference-time alignment via model collaboration . Preprint, arXiv:2410.09300

    2024 arXiv

  14. [22]

    Chelsea Finn, Pieter Abbeel, and Sergey Levine. 2017. Model-agnostic meta-learning for fast adaptation of deep networks. In International conference on machine learning, pages 1126--1135. PMLR

    2017

  15. [23]

    Junbo Fu, Guoshuai Zhao, Yimin Deng, Yunqi Mi, and Xueming Qian. 2024 a . https://doi.org/10.18653/v1/2024.findings-emnlp.134 Learning to paraphrase for alignment with LLM preference . In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 2394--2407, ...

    2024 doi

  16. [24]

    Tingchen Fu, Yupeng Hou, Julian McAuley, and Rui Yan. 2024 b . https://arxiv.org/abs/2408.05094 Unlocking decoding-time controllability: Gradient-free multi-objective alignment with contrastive prompts . Preprint, arXiv:2408.05094

    2024 arXiv

  17. [25]

    Iason Gabriel. 2020. https://doi.org/10.1007/s11023-020-09539-2 Artificial intelligence, values, and alignment . Minds and Machines, 30(3):411–437

    2020 doi

  18. [26]

    Jiahui Gao, Renjie Pi, Tianyang Han, Han Wu, Lanqing Hong, Lingpeng Kong, Xin Jiang, and Zhenguo Li. 2024 a . https://api.semanticscholar.org/CorpusID:272694191 Coca: Regaining safety-awareness of multimodal large language models with constitutional calibration . ArXiv, abs/2409.11365

    2024 arXiv

  19. [27]

    Songyang Gao, Qiming Ge, Wei Shen, Shihan Dou, Junjie Ye, Xiao Wang, Rui Zheng, Yicheng Zou, Zhi Chen, Hang Yan, Qi Zhang, and Dahua Lin. 2024 b . Linear alignment: a closed-form solution for aligning human preferences without tuning and feedback. In Proceedings of the 41st In...

    2024

  20. [28]

    Kwok, and Yu Zhang

    Yunhao Gou, Kai Chen, Zhili Liu, Lanqing Hong, Hang Xu, Zhenguo Li, Dit-Yan Yeung, James T. Kwok, and Yu Zhang. 2024. https://doi.org/10.1007/978-3-031-72643-9_23 Eyes closed, safety on: Protecting multimodal llms via image-to-text transformation . In Computer Vision – ECCV 20...

    2024 doi

  21. [29]

    Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, and Jacob Steinhardt. 2020. Measuring massive multitask language understanding. arXiv preprint arXiv:2009.03300

    2020 arXiv

  22. [30]

    Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. 2021. Lora: Low-rank adaptation of large language models. arXiv preprint arXiv:2106.09685

    2021 arXiv

  23. [31]

    Xiaomeng Hu, Pin-Yu Chen, and Tsung-Yi Ho. 2024. https://api.semanticscholar.org/CorpusID:274992169 Token highlighter: Inspecting and mitigating jailbreak prompts for large language models . In AAAI Conference on Artificial Intelligence

    2024

  24. [32]

    Huang, Sailik Sengupta, Daniele Bonadiman, Yi an Lai, Arshit Gupta, Nikolaos Pappas, Saab Mansour, Katrin Kirchhoff, and Dan Roth

    James Y. Huang, Sailik Sengupta, Daniele Bonadiman, Yi an Lai, Arshit Gupta, Nikolaos Pappas, Saab Mansour, Katrin Kirchhoff, and Dan Roth. 2024 a . https://arxiv.org/abs/2402.06147 Deal: Decoding-time alignment for large language models . Preprint, arXiv:2402.06147

    2024

  25. [33]

    Huang, Wenxuan Zhou, Fei Wang, Fred Morstatter, Sheng Zhang, Hoifung Poon, and Muhao Chen

    James Y. Huang, Wenxuan Zhou, Fei Wang, Fred Morstatter, Sheng Zhang, Hoifung Poon, and Muhao Chen. 2024 b . https://arxiv.org/abs/2404.11045 Offset unlearning for large language models . Preprint, arXiv:2404.11045

    2024 arXiv

  26. [34]

    Shih-Cheng Huang, Pin-Zu Li, Yu-chi Hsu, Kuang-Ming Chen, Yu Tung Lin, Shih-Kai Hsiao, Richard Tsai, and Hung-yi Lee. 2024 c . https://doi.org/10.18653/v1/2024.acl-long.590 Chat vector: A simple approach to equip LLM s with instruction following and model alignment in new lang...

    2024 doi

  27. [35]

    Yichong Huang, Baohang Li, Xiaocheng Feng, Wenshuai Huo, Chengpeng Fu, Ting Liu, and Bing Qin. 2024 d . https://doi.org/10.18653/v1/2024.emnlp-main.289 Aligning translation-specific understanding to general understanding in large language models . In Proceedings of the 2024 Co...

    2024 doi

  28. [36]

    Chia-Yu Hung, Navonil Majumder, Ambuj Mehrish, and Soujanya Poria. 2024. https://arxiv.org/abs/2406.15193 Inference time alignment with reward-guided tree search . Preprint, arXiv:2406.15193

    2024 arXiv

  29. [37]

    Aaron Hurst, Adam Lerer, Adam P Goucher, Adam Perelman, Aditya Ramesh, Aidan Clark, AJ Ostrow, Akila Welihinda, Alan Hayes, Alec Radford, et al. 2024. Gpt-4o system card. arXiv preprint arXiv:2410.21276

    2024 arXiv

  30. [38]

    Jiaming Ji, Boyuan Chen, Hantao Lou, Donghai Hong, Borong Zhang, Xuehai Pan, Juntao Dai, Tianyi Qiu, and Yaodong Yang. 2024 a . https://arxiv.org/abs/2402.02416 Aligner: Efficient alignment by learning to correct . Preprint, arXiv:2402.02416

    2024 arXiv

  31. [39]

    Jiaming Ji, Jiayi Zhou, Hantao Lou, Boyuan Chen, Donghai Hong, Xuyao Wang, Wenqi Chen, Kaile Wang, Rui Pan, Jiahao Li, Mohan Wang, Josef Dai, Tianyi Qiu, Hua Xu, Dong Li, Weipeng Chen, Jun Song, Bo Zheng, and Yaodong Yang. 2024 b . https://arxiv.org/abs/2412.15838 Align anythi...

    2024 arXiv

  32. [40]

    Ziwei Ji, Nayeon Lee, Rita Frieske, Tiezheng Yu, Dan Su, Yan Xu, Etsuko Ishii, Ye Jin Bang, Andrea Madotto, and Pascale Fung. 2023. https://doi.org/10.1145/3571730 Survey of hallucination in natural language generation . ACM Comput. Surv., 55(12)

    2023 doi

  33. [41]

    Yilei Jiang, Yingshui Tan, and Xiangyu Yue. 2024. https://arxiv.org/abs/2412.18826 Rapguard: Safeguarding multimodal large language models via rationale-aware defensive prompting . Preprint, arXiv:2412.18826

    2024 arXiv

  34. [42]

    Maxim Khanov, Jirayu Burapacheep, and Yixuan Li. 2024. https://api.semanticscholar.org/CorpusID:267411977 Args: Alignment as reward-guided search . ArXiv, abs/2402.01694

    2024 arXiv

  35. [43]

    Arsham Gholamzadeh Khoee, Yinan Yu, and Robert Feldt. 2024. Domain generalization through meta-learning: A survey. Artificial Intelligence Review, 57(10):285

    2024

  36. [44]

    Seanie Lee, Haebin Seong, Dong Bok Lee, Minki Kang, Xiaoyin Chen, Dominik Wagner, Yoshua Bengio, Juho Lee, and Sung Ju Hwang. 2024. https://arxiv.org/abs/2410.01524 Harmaug: Effective data augmentation for knowledge distillation of safety guard models . Preprint, arXiv:2410.01524

    2024 arXiv

  37. [45]

    Yongqi Leng and Deyi Xiong. 2025. https://arxiv.org/abs/2407.06488 Towards understanding multi-task learning (generalization) of llms via detecting and exploring task-specific neurons . Preprint, arXiv:2407.06488

    2025 arXiv

  38. [46]

    Chak Tou Leong, Yi Cheng, Jiashuo Wang, Jian Wang, and Wenjie Li. 2023. https://arxiv.org/abs/2310.09573 Self-detoxifying language models via toxification reversal . Preprint, arXiv:2310.09573

    2023 arXiv

  39. [47]

    Bolian Li, Yifan Wang, Anamika Lochab, Ananth Grama, and Ruqi Zhang. 2025 a . https://arxiv.org/abs/2406.16306 Cascade reward sampling for efficient decoding-time alignment . Preprint, arXiv:2406.16306

    2025 arXiv

  40. [48]

    Qing Li, Jiahui Geng, Derui Zhu, Zongxiong Chen, Kun Song, Lei Ma, and Fakhri Karray. 2025 b . Internal activation revision: Safeguarding vision language models without parameter update. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, pages 27428--27436

    2025

  41. [49]

    Yichuan Li, Kaize Ding, Jianling Wang, and Kyumin Lee. 2024 a . https://doi.org/10.18653/v1/2024.findings-acl.756 Empowering large language models for textual data augmentation . In Findings of the Association for Computational Linguistics: ACL 2024, pages 12734--12751, Bangko...

    2024 doi

  42. [50]

    Yuhui Li, Fangyun Wei, Jinjing Zhao, Chao Zhang, and Hongyang Zhang. 2024 b . https://openreview.net/forum?id=pETSfWMUzy RAIN : Your language models can align themselves without finetuning . In The Twelfth International Conference on Learning Representations

    2024

  43. [51]

    Bill Yuchen Lin, Abhilasha Ravichander, Ximing Lu, Nouha Dziri, Melanie Sclar, Khyathi Chandu, Chandra Bhagavatula, and Yejin Choi. 2023. The unlocking spell on base llms: Rethinking alignment via in-context learning. In The Twelfth International Conference on Learning Representations

    2023

  44. [52]

    Lin, Jacob Hilton, and Owain Evans

    Stephanie C. Lin, Jacob Hilton, and Owain Evans. 2021. https://api.semanticscholar.org/CorpusID:237532606 Truthfulqa: Measuring how models mimic human falsehoods . In Annual Meeting of the Association for Computational Linguistics

    2021

  45. [53]

    Alisa Liu, Xiaochuang Han, Yizhong Wang, Yulia Tsvetkov, Yejin Choi, and Noah A. Smith. 2024 a . https://arxiv.org/abs/2401.08565 Tuning language models by proxy . Preprint, arXiv:2401.08565

    2024 arXiv

  46. [54]

    Qin Liu, Chao Shang, Ling Liu, Nikolaos Pappas, Jie Ma, Neha Anna John, Srikanth Doss, Lluis Marquez, Miguel Ballesteros, and Yassine Benajiba. 2024 b . https://arxiv.org/abs/2410.09047 Unraveling and mitigating safety alignment degradation of vision-language models . Preprint...

    2024 arXiv

  47. [55]

    Qin Liu, Fei Wang, Chaowei Xiao, and Muhao Chen. 2025. https://arxiv.org/abs/2502.10486 Vlm-guard: Safeguarding vision-language models via fulfilling safety alignment gap . Preprint, arXiv:2502.10486

    2025 arXiv

  48. [56]

    Tianlin Liu, Shangmin Guo, Leonardo Bianco, Daniele Calandriello, Quentin Berthet, Felipe Llinares, Jessica Hoffmann, Lucas Dixon, Michal Valko, and Mathieu Blondel. 2024 c . https://arxiv.org/abs/2402.02992 Decoding-time realignment of language models . Preprint, arXiv:2402.02992

    2024 arXiv

  49. [57]

    Zhenyu Liu, Dongfang Li, Xinshuo Hu, Xinping Zhao, Yibin Chen, Baotian Hu, and Min Zhang. 2024 d . https://doi.org/10.18653/v1/2024.emnlp-main.160 Take off the training wheels! progressive in-context learning for effective alignment . In Proceedings of the 2024 Conference on E...

    2024 doi

  50. [58]

    Zhixuan Liu, Zhanhui Zhou, Yuanfu Wang, Chao Yang, and Yu Qiao. 2024 e . https://doi.org/10.18653/v1/2024.findings-emnlp.242 Inference-time language model alignment via integrated value guidance . In Findings of the Association for Computational Linguistics: EMNLP 2024, pages ...

    2024 doi

  51. [59]

    Daniel Machlab and Rick Battle. 2024. https://arxiv.org/abs/2404.08865 Llm in-context recall is prompt dependent . Preprint, arXiv:2404.08865

    2024 arXiv

  52. [60]

    Sewon Min, Xinxi Lyu, Ari Holtzman, Mikel Artetxe, Mike Lewis, Hannaneh Hajishirzi, and Luke Zettlemoyer. 2022. https://doi.org/10.18653/v1/2022.emnlp-main.759 Rethinking the role of demonstrations: What makes in-context learning work? In Proceedings of the 2022 Conference on ...

    2022 doi

  53. [61]

    Cohen, and Mirella Lapata

    Shashi Narayan, Shay B. Cohen, and Mirella Lapata. 2018. https://doi.org/10.18653/v1/D18-1206 Don`t give me the details, just the summary! topic-aware convolutional neural networks for extreme summarization . In Proceedings of the 2018 Conference on Empirical Methods in Natura...

    2018 doi

  54. [62]

    Long Ouyang, Jeff 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 Christiano, Jan Leike, and...

    2022

  55. [63]

    Mansi Phute, Alec Helbling, Matthew Hull, ShengYun Peng, Sebastian Szyller, Cory Cornelius, and Duen Horng Chau. 2024. https://arxiv.org/abs/2308.07308 Llm self defense: By self examination, llms know they are being tricked . Preprint, arXiv:2308.07308

    2024 arXiv

  56. [64]

    Chen Qian, Jie Zhang, Wei Yao, Dongrui Liu, Zhenfei Yin, Yu Qiao, Yong Liu, and Jing Shao. 2024. https://doi.org/10.18653/v1/2024.findings-acl.290 Towards tracing trustworthiness dynamics: Revisiting pre-training period of large language models . In Findings of the Association...

    2024 doi

  57. [65]

    Libo Qin, Qiguang Chen, Fuxuan Wei, Shijue Huang, and Wanxiang Che. 2023. https://doi.org/10.18653/v1/2023.emnlp-main.163 Cross-lingual prompting: Improving zero-shot chain-of-thought reasoning across languages . In Proceedings of the 2023 Conference on Empirical Methods in Na...

    2023 doi

  58. [66]

    Jiahao Qiu, Yifu Lu, Yifan Zeng, Jiacheng Guo, Jiayi Geng, Huazheng Wang, Kaixuan Huang, Yue Wu, and Mengdi Wang. 2024. https://arxiv.org/abs/2410.16033 Treebon: Enhancing inference-time alignment with speculative tree-search and best-of-n sampling . Preprint, arXiv:2410.16033

    2024 arXiv

  59. [67]

    Chris Richardson, Yao Zhang, Kellen Gillespie, Sudipta Kar, Arshdeep Singh, Zeynab Raeesy, Omar Zia Khan, and Abhinav Sethy. 2023. https://api.semanticscholar.org/CorpusID:264805263 Integrating summarization and retrieval for enhanced personalization via large language models ...

    2023 arXiv

  60. [68]

    Aadesh Salecha, Molly E Ireland, Shashanka Subrahmanya, Jo \ a o Sedoc, Lyle H Ungar, and Johannes C Eichstaedt. 2024. Large language models show human-like social desirability biases in survey responses. arXiv preprint arXiv:2405.06058

    2024 arXiv

  61. [69]

    Bilgehan Sel, Priya Shanmugasundaram, Mohammad Kachuee, Kun Zhou, Ruoxi Jia, and Ming Jin. 2024. https://doi.org/10.18653/v1/2024.acl-long.751 Skin-in-the-game: Decision making via multi-stakeholder alignment in LLM s . In Proceedings of the 62nd Annual Meeting of the Associat...

    2024 doi

  62. [70]

    C. E. Shannon. 1948. https://doi.org/10.1002/j.1538-7305.1948.tb01338.x A mathematical theory of communication . The Bell System Technical Journal, 27(3):379--423

    1948

  63. [71]

    Tianhao Shen, Renren Jin, Yufei Huang, Chuang Liu, Weilong Dong, Zishan Guo, Xinwei Wu, Yan Liu, and Deyi Xiong. 2023. https://arxiv.org/abs/2309.15025 Large language model alignment: A survey . Preprint, arXiv:2309.15025

    2023 arXiv

  64. [72]

    Freda Shi, Mirac Suzgun, Markus Freitag, Xuezhi Wang, Suraj Srivats, Soroush Vosoughi, Hyung Won Chung, Yi Tay, Sebastian Ruder, Denny Zhou, Dipanjan Das, and Jason Wei. 2022. https://arxiv.org/abs/2210.03057 Language models are multilingual chain-of-thought reasoners . Prepri...

    2022 arXiv

  65. [73]

    Ruizhe Shi, Yifang Chen, Yushi Hu, Alisa Liu, Hannaneh Hajishirzi, Noah A Smith, and Simon S Du. 2024. Decoding-time language model alignment with multiple objectives. arXiv preprint arXiv:2406.18853

    2024 arXiv

  66. [74]

    Chenglei Si, Zhe Gan, Zhengyuan Yang, Shuohang Wang, Jianfeng Wang, Jordan Lee Boyd-Graber, and Lijuan Wang. 2023. https://openreview.net/forum?id=98p5x51L5af Prompting GPT -3 to be reliable . In The Eleventh International Conference on Learning Representations

    2023

  67. [75]

    Soice, Rafael Rocha, Kimberlee Cordova, Michael Specter, and Kevin M

    Emily H. Soice, Rafael Rocha, Kimberlee Cordova, Michael Specter, and Kevin M. Esvelt. 2023. https://arxiv.org/abs/2306.03809 Can large language models democratize access to dual-use biotechnology? Preprint, arXiv:2306.03809

    2023 arXiv

  68. [76]

    Feifan Song, Yuxuan Fan, Xin Zhang, Peiyi Wang, and Houfeng Wang. 2025 a . https://doi.org/10.18653/v1/2025.naacl-long.8 Instantly learning preference alignment via in-context DPO . In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association...

    2025 doi

  69. [77]

    Feifan Song, Bofei Gao, Yifan Song, Yi Liu, Weimin Xiong, Yuyang Song, Tianyu Liu, Guoyin Wang, and Houfeng Wang. 2025 b . https://arxiv.org/abs/2508.04626 P-aligner: Enabling pre-alignment of language models via principled instruction synthesis . Preprint, arXiv:2508.04626

    2025 arXiv

  70. [78]

    Feifan Song, Shaohang Wei, Wen Luo, Yuxuan Fan, Tianyu Liu, Guoyin Wang, and Houfeng Wang. 2025 c . https://doi.org/10.18653/v1/2025.findings-acl.655 Well begun is half done: Low-resource preference alignment by weak-to-strong decoding . In Findings of the Association for Comp...

    2025 doi

  71. [79]

    Robin Staab, Mark Vero, Mislav Balunovic, and Martin Vechev. 2024. https://openreview.net/forum?id=kmn0BhQk7p Beyond memorization: Violating privacy via inference with large language models . In The Twelfth International Conference on Learning Representations

    2024

  72. [80]

    Tianyi Tang, Wenyang Luo, Haoyang Huang, Dongdong Zhang, Xiaolei Wang, Xin Zhao, Furu Wei, and Ji-Rong Wen. 2024. https://doi.org/10.18653/v1/2024.acl-long.309 Language-specific neurons: The key to multilingual capabilities in large language models . In Proceedings of the 62nd...

    2024 doi

  73. [81]

    Eshaan Tanwar, Manish Borthakur, Subhabrata Dutta, and Tanmoy Chakraborty. 2023. https://api.semanticscholar.org/CorpusID:258588286 Multilingual llms are better cross-lingual in-context learners with alignment . ArXiv, abs/2305.05940

    2023 arXiv

  74. [82]

    Chameleon Team. 2024. https://arxiv.org/abs/2405.09818 Chameleon: Mixed-modal early-fusion foundation models . Preprint, arXiv:2405.09818

    2024 arXiv

  75. [83]

    Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timoth \'e e Lacroix, Baptiste Rozi \`e re, Naman Goyal, Eric Hambro, Faisal Azhar, et al. 2023. Llama: Open and efficient foundation language models. arXiv preprint arXiv:2302.13971

    2023 arXiv

  76. [84]

    Mengru Wang, Ningyu Zhang, Ziwen Xu, Zekun Xi, Shumin Deng, Yunzhi Yao, Qishen Zhang, Linyi Yang, Jindong Wang, and Huajun Chen. 2024 a . https://doi.org/10.18653/v1/2024.acl-long.171 Detoxifying large language models via knowledge editing . In Proceedings of the 62nd Annual M...

    2024 doi

  77. [85]

    Pengyu Wang, Dong Zhang, Linyang Li, Chenkun Tan, Xinghao Wang, Mozhi Zhang, Ke Ren, Botian Jiang, and Xipeng Qiu. 2024 b . https://doi.org/10.18653/v1/2024.emnlp-main.585 I nfer A ligner: Inference-time alignment for harmlessness through cross-model guidance . In Proceedings ...

    2024 doi

  78. [86]

    Yu Wang, Xiaogeng Liu, Yu Li, Muhao Chen, and Chaowei Xiao. 2024 c . https://arxiv.org/abs/2403.09513 Adashield: Safeguarding multimodal large language models from structure-based attack via adaptive shield prompting . Preprint, arXiv:2403.09513

    2024 arXiv

  79. [87]

    Zhichao Wang, Bin Bi, Shiva Kumar Pentyala, Kiran Ramnath, Sougata Chaudhuri, Shubham Mehrotra, Zixu, Zhu, Xiang-Bo Mao, Sitaram Asur, Na, and Cheng. 2024 d . https://arxiv.org/abs/2407.16216 A comprehensive survey of llm alignment techniques: Rlhf, rlaif, ppo, dpo and more . ...

    2024 arXiv

  80. [88]

    Zeming Wei, Yifei Wang, Ang Li, Yichuan Mo, and Yisen Wang. 2024. https://arxiv.org/abs/2310.06387 Jailbreak and guard aligned language models with only few in-context demonstrations . Preprint, arXiv:2310.06387

    2024 arXiv

  81. [89]

    Xuansheng Wu, Haiyan Zhao, Yaochen Zhu, Yucheng Shi, Fan Yang, Tianming Liu, Xiaoming Zhai, Wenlin Yao, Jundong Li, Mengnan Du, and Ninghao Liu. 2024. https://arxiv.org/abs/2403.08946 Usable xai: 10 strategies towards exploiting explainability in the llm era . Preprint, arXiv:...

    2024 arXiv

  82. [90]

    Jinheng Xie, Weijia Mao, Zechen Bai, David Junhao Zhang, Weihao Wang, Kevin Qinghong Lin, Yuchao Gu, Zhijie Chen, Zhenheng Yang, and Mike Zheng Shou. 2024. https://arxiv.org/abs/2408.12528 Show-o: One single transformer to unify multimodal understanding and generation . Prepri...

    2024 arXiv

  83. [91]

    Yueqi Xie, Jingwei Yi, Jiawei Shao, Justin Curl, Lingjuan Lyu, Qifeng Chen, Xing Xie, and Fangzhao Wu. 2023. Defending chatgpt against jailbreak attack via self-reminders. Nature Machine Intelligence, 5(12):1486--1496

    2023

  84. [92]

    Jing Xiong, Gongye Liu, Lun Huang, Chengyue Wu, Taiqiang Wu, Yao Mu, Yuan Yao, Hui Shen, Zhongwei Wan, Jinfa Huang, Chaofan Tao, Shen Yan, Huaxiu Yao, Lingpeng Kong, Hongxia Yang, Mi Zhang, Guillermo Sapiro, Jiebo Luo, Ping Luo, and Ngai Wong. 2024. https://arxiv.org/abs/2411....

    2024 arXiv

  85. [93]

    Chunpu Xu, Steffi Chern, Ethan Chern, Ge Zhang, Zekun Wang, Ruibo Liu, Jing Li, Jie Fu, and Pengfei Liu. 2023. https://arxiv.org/abs/2312.15907 Align on the fly: Adapting chatbot behavior to established norms . Preprint, arXiv:2312.15907

    2023 arXiv

  86. [94]

    Yuancheng Xu, Udari Madhushani Sehwag, Alec Koppel, Sicheng Zhu, Bang An, Furong Huang, and Sumitra Ganesh. 2025. https://arxiv.org/abs/2410.08193 Genarm: Reward guided generation with autoregressive reward model for test-time alignment . Preprint, arXiv:2410.08193

    2025 arXiv

  87. [95]

    Yue Xu, Xiuyuan Qi, Zhan Qin, and Wenjie Wang. 2024 a . https://doi.org/10.18653/v1/2024.findings-emnlp.803 Cross-modality information check for detecting jailbreaking in multimodal large language models . In Findings of the Association for Computational Linguistics: EMNLP 202...

    2024 doi

  88. [96]

    Lin, and Radha Poovendran

    Zhangchen Xu, Fengqing Jiang, Niu Luyao, Jia Jinyuan, Bill Y. Lin, and Radha Poovendran. 2024 b . https://arxiv.org/abs/2402.08983 Safedecoding: Defending against jailbreak attacks via safety-aware decoding . In Annual Meeting of the Association for Computational Linguistics

    2024 arXiv

  89. [97]

    Qinghao Ye, Haiyang Xu, Jiabo Ye, Ming Yan, Anwen Hu, Haowei Liu, Qi Qian, Ji Zhang, and Fei Huang. 2024. mplug-owl2: Revolutionizing multi-modal large language model with modality collaboration. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognit...

    2024

  90. [98]

    Xinyi Zeng, Yuying Shang, Jiawei Chen, Jingyuan Zhang, and Yu Tian. 2025. https://arxiv.org/abs/2410.06809 Root defence strategies: Ensuring safety of llm at the decoding level . Preprint, arXiv:2410.06809

    2025 arXiv

  91. [99]

    Runzhe Zhan, Xinyi Yang, Derek Wong, Lidia Chao, and Yue Zhang. 2024. https://doi.org/10.18653/v1/2024.findings-acl.722 Prefix text as a yarn: Eliciting non- E nglish alignment in foundation language model . In Findings of the Association for Computational Linguistics: ACL 202...

    2024 doi

  92. [100]

    Jingyu Zhang, Ahmed Elgohary, Ahmed Magooda, Daniel Khashabi, and Benjamin Van Durme. 2024 a . https://arxiv.org/abs/2410.08968 Controllable safety alignment: Inference-time adaptation to diverse safety requirements . Preprint, arXiv:2410.08968

    2024 arXiv

  93. [101]

    LingXi Zhang, Yue Yu, Kuan Wang, and Chao Zhang. 2024 b . https://doi.org/10.18653/v1/2024.acl-long.203 ARL 2: Aligning retrievers with black-box large language models via self-guided adaptive relevance labeling . In Proceedings of the 62nd Annual Meeting of the Association fo...

    2024 doi

  94. [102]

    Shuo Zhang, Liangming Pan, Junzhou Zhao, and William Yang Wang. 2024 c . https://doi.org/10.18653/v1/2024.findings-acl.121 The knowledge alignment problem: Bridging human and external knowledge for large language models . In Findings of the Association for Computational Lingui...

    2024 doi

  95. [103]

    Yongheng Zhang, Qiguang Chen, Min Li, Wanxiang Che, and Libo Qin. 2024 d . https://doi.org/10.18653/v1/2024.findings-acl.546 A uto CAP : Towards automatic cross-lingual alignment planning for zero-shot chain-of-thought . In Findings of the Association for Computational Linguis...

    2024 doi

  96. [104]

    Zhexin Zhang, Junxiao Yang, Pei Ke, Fei Mi, Hongning Wang, and Minlie Huang. 2024 e . https://doi.org/10.18653/v1/2024.acl-long.481 Defending large language models against jailbreaking attacks through goal prioritization . In Proceedings of the 62nd Annual Meeting of the Assoc...

    2024 doi

  97. [105]

    Zihan Zhang, Meng Fang, Ling Chen, Mohammad-Reza Namazi-Rad, and Jun Wang. 2023. https://doi.org/10.18653/v1/2023.emnlp-main.516 How do large language models capture the ever-changing world knowledge? a review of recent advances . In Proceedings of the 2023 Conference on Empir...

    2023 doi

  98. [106]

    Xueliang Zhao, Xinting Huang, Tingchen Fu, Qintong Li, Shansan Gong, Lemao Liu, Wei Bi, and Lingpeng Kong. 2024. https://doi.org/10.18653/v1/2024.findings-acl.433 BBA : Bi-modal behavioral alignment for reasoning with large vision-language models . In Findings of the Associati...

    2024 doi

  99. [107]

    Chunting Zhou, Pengfei Liu, Puxin Xu, Srinivasan Iyer, Jiao Sun, Yuning Mao, Xuezhe Ma, Avia Efrat, Ping Yu, Lili Yu, et al. 2023. Lima: Less is more for alignment. Advances in Neural Information Processing Systems, 36:55006--55021

    2023

  100. [108]

    Chunting Zhou, Lili Yu, Arun Babu, Kushal Tirumala, Michihiro Yasunaga, Leonid Shamis, Jacob Kahn, Xuezhe Ma, Luke Zettlemoyer, and Omer Levy. 2024 a . https://arxiv.org/abs/2408.11039 Transfusion: Predict the next token and diffuse images with one multi-modal model . Preprint...

    2024 arXiv

  101. [109]

    Qi Zhou, Tianlin Li, Qing Guo, Dongxia Wang, Yun Lin, Yang Liu, and Jin Song Dong. 2024 b . https://arxiv.org/abs/2412.12722 Defending lvlms against vision attacks through partial-perception supervision . Preprint, arXiv:2412.12722

    2024 arXiv

  102. [110]

    Mingye Zhu, Yi Liu, Lei Zhang, Junbo Guo, and Zhendong Mao. 2025. https://openreview.net/forum?id=cfn2O1qvxp On-the-fly preference alignment via principle-guided decoding . In The Thirteenth International Conference on Learning Representations

    2025

  103. [111]

    Minjun Zhu, Linyi Yang, and Yue Zhang. 2024. https://arxiv.org/abs/2408.11779 Personality alignment of large language models . Preprint, arXiv:2408.11779

    2024 arXiv

  104. [112]

    Yuchen Zhuang, Haotian Sun, Yue Yu, Rushi Qiang, Qifan Wang, Chao Zhang, and Bo Dai. 2024. Hydra: Model factorization framework for black-box llm personalization. arXiv preprint arXiv:2406.02888

    2024 arXiv

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.