pith. machine review for the scientific record. sign in

arxiv: 2604.24536 · v1 · submitted 2026-04-27 · 💻 cs.CL

Recognition: unknown

Generating Place-Based Compromises Between Two Points of View

Charlene C. Wu, Francine Chen, Kate Sieck, Monica P. Van, Nayeli Suseth Bravo, Scott Carter, Sumanta Bhattacharyya, Tatiana Lau, Yan-Ying Chen

Authors on Pith no claims yet

Pith reviewed 2026-05-08 03:32 UTC · model grok-4.3

classification 💻 cs.CL
keywords compromise generationempathic neutralityprompt engineeringlarge language modelshuman evaluationmodel alignmentviewpoint reconciliation
0
0 comments X

The pith

Using iterative empathic similarity feedback generates more acceptable compromises between opposing viewpoints than standard chain-of-thought reasoning.

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

This paper explores ways to make large language models better at creating compromises between two conflicting opinions about a place. It tests different prompting techniques and discovers that repeatedly checking how similar the compromise feels to each side using empathic measures leads to superior outcomes. Human participants in a study found these compromises more acceptable compared to those from basic reasoning prompts. The generated compromises are then used to train smaller models that perform well without needing the same feedback process during use.

Core claim

The authors demonstrate that an iterative process incorporating external assessments of empathic similarity to both viewpoints produces compromises with higher acceptability ratings in a human evaluation study than those generated via standard chain-of-thought prompting, and that the resulting dataset supports training smaller models through margin-based preference alignment for efficient deployment.

What carries the argument

The iterative empathic similarity feedback mechanism, which evaluates proposed compromises for their neutrality by measuring similarity to each original viewpoint and uses this to guide further generation.

If this is right

  • Compromises from the empathic feedback method receive higher acceptability scores from human evaluators than those from chain-of-thought.
  • Smaller models trained on the generated dataset achieve comparable performance while eliminating the need for empathy estimation at inference.
  • The method improves the social intelligence of LLMs in handling viewpoint conflicts on shared places.
  • Margin-based alignment allows efficient transfer of the compromise generation capability to compact models.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Similar feedback techniques might enhance compromise generation in non-place-based disputes if appropriate similarity metrics are defined.
  • Integration into dialogue systems could help mediate online disagreements in real time using the trained smaller models.
  • Expanding the dataset with more diverse viewpoints could reduce potential biases in the trained models.

Load-bearing premise

The 50-participant acceptability study accurately measures compromise quality without bias from the specific participants or the presentation of the viewpoints.

What would settle it

Conducting an acceptability study with a much larger and demographically diverse participant group that shows no advantage for the empathic similarity method over standard chain-of-thought prompting.

Figures

Figures reproduced from arXiv: 2604.24536 by Charlene C. Wu, Francine Chen, Kate Sieck, Monica P. Van, Nayeli Suseth Bravo, Scott Carter, Sumanta Bhattacharyya, Tatiana Lau, Yan-Ying Chen.

Figure 1
Figure 1. Figure 1: Example showing two contrasting views, View_A and View_B, two compromises generated for the two view at source ↗
Figure 2
Figure 2. Figure 2: Prompt engineering strategies for collected compromise generation (a) basic CoT approach (b) CoT+LLM view at source ↗
Figure 3
Figure 3. Figure 3: Overview of the full pipeline used in our proposed method. view at source ↗
Figure 4
Figure 4. Figure 4: Difference between score_A and score_B: lower values indicate higher neutrality. The baseline model (Mistral-7b) produces less neutral compromises, resulting in a higher score difference. In contrast, the compromises generated via iterative prompt engineer￾ing achieve the lowest difference, reflecting the highest neutrality. After aligning the base model, it produces significantly more neutral compromises … view at source ↗
Figure 5
Figure 5. Figure 5: Data collection methodology D Collection of Human Ratings for Evaluation of Prompting Strategies The Prolific3 platform was used to recruit partici￾pants located in the US. The participants were paid the equivalent of $17/hr for 20 minutes of time, with the actual mean time of a batch of participants ranging from 14-20 minutes. All participants pro￾vided informed consent prior to participating in the study… view at source ↗
read the original abstract

Large Language Models (LLMs) excel academically but struggle with social intelligence tasks, such as creating good compromises. In this paper, we present methods for generating empathically neutral compromises between two opposing viewpoints. We first compared four different prompt engineering methods using Claude 3 Opus and a dataset of 2,400 contrasting views on shared places. A subset of the gen erated compromises was evaluated for acceptability in a 50-participant study. We found that the best method for generating compromises between two views used external empathic similarity between a compromise and each viewpoint as iterative feedback, outperforming stan dard Chain of Thought (CoT) reasoning. The results indicate that the use of empathic neutrality improves the acceptability of compromises. The dataset of generated compromises was then used to train two smaller foundation models via margin-based alignment of human preferences, improving efficiency and removing the need for empathy estimation during inference.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper proposes prompt-engineering methods for generating empathically neutral compromises between opposing viewpoints on shared places. Using Claude 3 Opus and a dataset of 2,400 contrasting views, it compares four methods and identifies iterative external empathic similarity feedback as superior to standard Chain-of-Thought reasoning. A 50-participant acceptability study supports the claim that empathic neutrality improves compromise quality. The generated dataset is then used to train smaller models via margin-based alignment of human preferences, enabling efficient inference without ongoing empathy estimation.

Significance. If the empirical findings hold after proper validation, the work offers a concrete technique for enhancing LLMs on social-intelligence tasks such as compromise generation. The distillation step to smaller models is a practical strength that could improve deployability. The iterative empathic-feedback approach is a potentially reusable idea for prompt design in value-laden domains.

major comments (2)
  1. [Evaluation section (and abstract)] The 50-participant acceptability study is load-bearing for the central claim that the empathic method outperforms CoT and improves acceptability, yet the manuscript provides no details on participant recruitment, blinding, inter-rater agreement, exact scoring protocol, or statistical tests (e.g., no mention of p-values, effect sizes, or power analysis).
  2. [Model training and alignment section] Training the smaller foundation models on compromises generated by the same empathic method under evaluation creates a circularity risk: any systematic artifacts or biases in the original LLM outputs are inherited by the fine-tuned models rather than validated against independent human or external benchmarks.
minor comments (2)
  1. [Abstract] Abstract contains typographical errors (e.g., 'gen erated', 'stan dard') that should be corrected for clarity.
  2. [Methods] The description of the four prompt-engineering methods would benefit from explicit pseudocode or example prompts to allow replication.

Simulated Author's Rebuttal

2 responses · 0 unresolved

Thank you for the constructive feedback. We have carefully considered the major comments and provide point-by-point responses below. We will revise the manuscript accordingly to address the concerns raised.

read point-by-point responses

  1. Referee: The 50-participant acceptability study is load-bearing for the central claim that the empathic method outperforms CoT and improves acceptability, yet the manuscript provides no details on participant recruitment, blinding, inter-rater agreement, exact scoring protocol, or statistical tests (e.g., no mention of p-values, effect sizes, or power analysis).

    Authors: We agree that additional details on the human evaluation are necessary for reproducibility and to support the claims. In the revised manuscript, we will expand the Evaluation section to include: participant recruitment method and demographics, blinding procedures (if any), inter-rater agreement statistics, the precise scoring protocol used, and results of statistical tests including p-values and effect sizes. We will also include a power analysis if feasible based on the collected data. This will be added without altering the core findings. revision: yes

  2. Referee: Training the smaller foundation models on compromises generated by the same empathic method under evaluation creates a circularity risk: any systematic artifacts or biases in the original LLM outputs are inherited by the fine-tuned models rather than validated against independent human or external benchmarks.

    Authors: This is a valid concern regarding potential propagation of biases from the generator LLM. However, the alignment process relies on human preference judgments collected on the generated compromises, using a margin-based loss to train the smaller models. These human preferences serve as an independent signal. We will revise the Model training and alignment section to explicitly describe the human annotation process, clarify how it mitigates circularity, and add a discussion of limitations, including the need for future work with human-generated compromises as benchmarks. revision: partial

Circularity Check

0 steps flagged

No significant circularity; evaluation rests on independent human study

full rationale

The paper's core result—that external empathic similarity feedback outperforms standard CoT—is established by comparing four prompt methods on 2,400 view pairs and then measuring acceptability via a separate 50-participant human study. This human rating serves as an external benchmark rather than a self-referential metric. The subsequent step of training smaller models on the generated dataset via margin-based alignment of human preferences is a downstream distillation task; it does not redefine or presuppose the superiority claim, nor does it create a fitted-input-called-prediction loop. No self-citations, uniqueness theorems, or definitional equivalences appear in the described chain, so the derivation remains self-contained against the external human evaluation.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract supplies no explicit free parameters, axioms, or invented entities. The method implicitly assumes that empathic similarity can be computed reliably by an external process and that margin-based preference alignment transfers the desired behavior to smaller models.

pith-pipeline@v0.9.0 · 5478 in / 1245 out tokens · 82021 ms · 2026-05-08T03:32:04.484876+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

81 extracted references · 37 canonical work pages · 8 internal anchors

  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]

    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

    work page internal anchor Pith review arXiv 2023

  4. [4]

    AI Anthropic. 2023. Introducing claude

    2023

  5. [5]

    Mohammad Gheshlaghi Azar, Zhaohan Daniel Guo, Bilal Piot, Remi Munos, Mark Rowland, Michal Valko, and Daniele Calandriello. 2024. A general theoretical paradigm to understand learning from human preferences. In International Conference on Artificial Intelligence and Statistics, pages 4447--4455. PMLR

    2024

  6. [6]

    Michiel Bakker, Martin Chadwick, Hannah Sheahan, Michael Tessler, Lucy Campbell-Gillingham, Jan Balaguer, Nat McAleese, Amelia Glaese, John Aslanides, Matt Botvinick, et al. 2022. Fine-tuning language models to find agreement among humans with diverse preferences. Advances in Neural Information Processing Systems, 35:38176--38189

    2022

  7. [7]

    Krisztian Balog, John Palowitch, Barbara Ikica, Filip Radlinski, Hamidreza Alvari, and Mehdi Manshadi. 2024. Towards realistic synthetic user-generated content: A scaffolding approach to generating online discussions. CoRR

    2024

  8. [8]

    Cristina Bicchieri. 2005. The grammar of society: The nature and dynamics of social norms. Cambridge University Press

    2005

  9. [9]

    Paola Cascante-Bonilla, Khaled Shehada, James Seale Smith, Sivan Doveh, Donghyun Kim, Rameswar Panda, Gul Varol, Aude Oliva, Vicente Ordonez, Rogerio Feris, et al. 2023. Going beyond nouns with vision & language models using synthetic data. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 20155--20165

    2023

  10. [10]

    Francine Chen, Scott Carter, Tatiana Lau, Nayeli Suseth Bravo, Sumanta Bhattacharyya, Kate Sieck, and Charlene C Wu. 2025. Empathy prediction from diverse perspectives. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 8959--8974

    2025

  11. [11]

    Huayu Chen, Guande He, Lifan Yuan, Ganqu Cui, Hang Su, and Jun Zhu. 2024. Noise contrastive alignment of language models with explicit rewards. arXiv preprint arXiv:2402.05369

    work page arXiv 2024

  12. [12]

    Minje Choi, Jiaxin Pei, Sagar Kumar, Chang Shu, and David Jurgens. 2023. Do llms understand social knowledge? evaluating the sociability of large language models with socket benchmark. arXiv preprint arXiv:2305.14938

    work page arXiv 2023

  13. [13]

    Zheng Chu, Jingchang Chen, Qianglong Chen, Weijiang Yu, Tao He, Haotian Wang, Weihua Peng, Ming Liu, Bing Qin, and Ting Liu. 2023. A survey of chain of thought reasoning: Advances, frontiers and future. arXiv preprint arXiv:2309.15402

    work page arXiv 2023

  14. [14]

    Zheng Chu, Jingchang Chen, Qianglong Chen, Weijiang Yu, Tao He, Haotian Wang, Weihua Peng, Ming Liu, Bing Qin, and Ting Liu. 2024. Navigate through enigmatic labyrinth a survey of chain of thought reasoning: Advances, frontiers and future. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), p...

    2024

  15. [15]

    Fida K Dankar and Mahmoud Ibrahim. 2021. Fake it till you make it: Guidelines for effective synthetic data generation. Applied Sciences, 11(5):2158

    2021

  16. [16]

    Giulia DeSalvo, Jean-Fracois Kagy, Lazaros Karydas, Afshin Rostamizadeh, and Sanjiv Kumar. 2024. No more hard prompts: Softsrv prompting for synthetic data generation. arXiv preprint arXiv:2410.16534

    work page arXiv 2024

  17. [17]

    Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, et al. 2024. The llama 3 herd of models. arXiv preprint arXiv:2407.21783

    work page internal anchor Pith review arXiv 2024

  18. [18]

    Nicholas Epley and Thomas Gilovich. 2006. The anchoring-and-adjustment heuristic: Why the adjustments are insufficient. Psychological science, 17(4):311--318

    2006

  19. [19]

    Muhammad Fawi. 2024. Curlora: Stable llm continual fine-tuning and catastrophic forgetting mitigation. arXiv preprint arXiv:2408.14572

    work page arXiv 2024

  20. [20]

    Ernst Fehr and Urs Fischbacher. 2004. Social norms and human cooperation. Trends in cognitive sciences, 8(4):185--190

    2004

  21. [21]

    Guhao Feng, Bohang Zhang, Yuntian Gu, Haotian Ye, Di He, and Liwei Wang. 2024. Towards revealing the mystery behind chain of thought: a theoretical perspective. Advances in Neural Information Processing Systems, 36

    2024

  22. [22]

    Adam Fisch, Jacob Eisenstein, Vicky Zayats, Alekh Agarwal, Ahmad Beirami, Chirag Nagpal, Pete Shaw, and Jonathan Berant. 2024. Robust preference optimization through reward model distillation. arXiv preprint arXiv:2405.19316

    work page arXiv 2024

  23. [23]

    Rodney A Gabriel, Onkar Litake, Sierra Simpson, Brittany N Burton, Ruth S Waterman, and Alvaro A Macias. 2024. On the development and validation of large language model-based classifiers for identifying social determinants of health. Proceedings of the National Academy of Sciences, 121(39):e2320716121

    2024

  24. [24]

    Hamideh Ghanadian, Isar Nejadgholi, and Hussein Al Osman. 2024. Socially aware synthetic data generation for suicidal ideation detection using large language models. IEEE Access

    2024

  25. [25]

    Zhiqiang Gong, Ping Zhong, and Weidong Hu. 2019. Diversity in machine learning. Ieee Access, 7:64323--64350

    2019

  26. [26]

    Bhanu Prakash Reddy Guda, Aparna Garimella, and Niyati Chhaya. 2021. Empathbert: A bert-based framework for demographic-aware empathy prediction. arXiv preprint arXiv:2102.00272

    work page arXiv 2021

  27. [27]

    Yuxiang Guo, Lu Yin, Bo Jiang, and Jiaqi Zhang. 2024. Todo: Enhancing llm alignment with ternary preferences. arXiv preprint arXiv:2411.02442

    work page arXiv 2024

  28. [28]

    Michael Gutmann and Aapo Hyv \"a rinen. 2010. Noise-contrastive estimation: A new estimation principle for unnormalized statistical models. In Proceedings of the thirteenth international conference on artificial intelligence and statistics, pages 297--304. JMLR Workshop and Conference Proceedings

    2010

  29. [29]

    Michael U Gutmann and Aapo Hyv \"a rinen. 2012. Noise-contrastive estimation of unnormalized statistical models, with applications to natural image statistics. Journal of machine learning research, 13(2)

    2012

  30. [30]

    Md Rakibul Hasan, Md Zakir Hossain, Tom Gedeon, and Shafin Rahman. 2024. Llm-gem: Large language model-guided prediction of people’s empathy levels towards newspaper article. In Findings of the Association for Computational Linguistics: EACL 2024, pages 2215--2231

    2024

  31. [31]

    Abdelrahaman A Hassan, Radwa J Hanafy, and Mohammed E Fouda. 2024. Automated multi-label annotation for mental health illnesses using large language models. arXiv preprint arXiv:2412.03796

    work page arXiv 2024

  32. [32]

    Asen Hikov and Laura Murphy. 2024. Information retrieval from textual data: Harnessing large language models, retrieval augmented generation and prompt engineering. Journal of AI, Robotics & Workplace Automation, 3(2):142--150

    2024

  33. [33]

    Jixiang Hong, Quan Tu, Changyu Chen, Xing Gao, Ji Zhang, and Rui Yan. 2023. Cyclealign: Iterative distillation from black-box llm to white-box models for better human alignment. arXiv preprint arXiv:2310.16271

    work page arXiv 2023

  34. [34]

    Guiyang Hou, Wenqi Zhang, Yongliang Shen, Zeqi Tan, Sihao Shen, and Weiming Lu. 2024. Entering real social world! benchmarking the theory of mind and socialization capabilities of llms from a first-person perspective. arXiv preprint arXiv:2410.06195

    work page arXiv 2024

  35. [35]

    Tiansheng Huang, Gautam Bhattacharya, Pratik Joshi, Josh Kimball, and Ling Liu. 2024. Antidote: Post-fine-tuning safety alignment for large language models against harmful fine-tuning. arXiv preprint arXiv:2408.09600

    work page arXiv 2024

  36. [36]

    Dieuwke Hupkes, Mario Giulianelli, Verna Dankers, Mikel Artetxe, Yanai Elazar, Tiago Pimentel, Christos Christodoulopoulos, Karim Lasri, Naomi Saphra, Arabella Sinclair, et al. 2022. State-of-the-art generalisation research in nlp: a taxonomy and review. arXiv preprint arXiv:2210.03050

    work page arXiv 2022

  37. [37]

    Albert Q Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lucile Saulnier, et al. 2023. Mistral 7b. arXiv preprint arXiv:2310.06825

    work page internal anchor Pith review arXiv 2023

  38. [38]

    Timo Kaufmann, Paul Weng, Viktor Bengs, and Eyke H \"u llermeier. 2023. A survey of reinforcement learning from human feedback. arXiv preprint arXiv:2312.14925

    work page arXiv 2023

  39. [39]

    Dongyoung Kim, Kimin Lee, Jinwoo Shin, and Jaehyung Kim. 2024. Aligning large language models with self-generated preference data. arXiv preprint arXiv:2406.04412

    work page arXiv 2024

  40. [40]

    Solomon Kullback and Richard A Leibler. 1951. On information and sufficiency. The annals of mathematical statistics, 22(1):79--86

    1951

  41. [41]

    Matthew Le, Y-Lan Boureau, and Maximilian Nickel. 2019. Revisiting the evaluation of theory of mind through question answering. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 5872--5877

    2019

  42. [42]

    Fangyu Lei, Qian Liu, Yiming Huang, Shizhu He, Jun Zhao, and Kang Liu. 2023. S3eval: A synthetic, scalable, systematic evaluation suite for large language models. arXiv preprint arXiv:2310.15147

    work page arXiv 2023

  43. [43]

    Chin-Yew Lin. 2004. Rouge: A package for automatic evaluation of summaries. In Text summarization branches out, pages 74--81

    2004

  44. [44]

    Wenhao Liu, Xiaohua Wang, Muling Wu, Tianlong Li, Changze Lv, Zixuan Ling, Jianhao Zhu, Cenyuan Zhang, Xiaoqing Zheng, and Xuanjing Huang. 2023. Aligning large language models with human preferences through representation engineering. arXiv preprint arXiv:2312.15997

    work page arXiv 2023

  45. [45]

    Ziyi Liu, Abhishek Anand, Pei Zhou, Jen-tse Huang, and Jieyu Zhao. 2024. Interintent: Investigating social intelligence of llms via intention understanding in an interactive game context. arXiv preprint arXiv:2406.12203

    work page arXiv 2024

  46. [46]

    Zilin Ma, Nathan Zhao, Linn Bieske, Blake Bullwinkel, Yanyi Zhang, Ziqing Luo, Siyao Li, Gekai Liao, Boxiang Wang, Jinglun Gao, et al. 2024. Using large language models for humanitarian frontline negotiation: Opportunities and considerations. arXiv preprint arXiv:2405.20195

    work page arXiv 2024

  47. [47]

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

    2024

  48. [48]

    Navonil Majumder, Pengfei Hong, Shanshan Peng, Jiankun Lu, Deepanway Ghosal, Alexander Gelbukh, Rada Mihalcea, and Soujanya Poria. 2020. Mime: Mimicking emotions for empathetic response generation. arXiv preprint arXiv:2010.01454

    work page arXiv 2020

  49. [49]

    Stephen Merity, Nitish Shirish Keskar, James Bradbury, and Richard Socher. 2018. Scalable language modeling: Wikitext-103 on a single gpu in 12 hours. Proceedings of the SYSML, 18

    2018

  50. [50]

    Yu Nong, Mohammed Aldeen, Long Cheng, Hongxin Hu, Feng Chen, and Haipeng Cai. 2024. Chain-of-thought prompting of large language models for discovering and fixing software vulnerabilities. arXiv preprint arXiv:2402.17230

    work page arXiv 2024

  51. [51]

    Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. 2022. Training language models to follow instructions with human feedback. Advances in neural information processing systems, 35:27730--27744

    2022

  52. [52]

    Arka Pal, Deep Karkhanis, Samuel Dooley, Manley Roberts, Siddartha Naidu, and Colin White. 2024. Smaug: Fixing failure modes of preference optimisation with dpo-positive. arXiv preprint arXiv:2402.13228

    work page arXiv 2024

  53. [53]

    Rafael Rafailov, Archit Sharma, Eric Mitchell, Christopher D Manning, Stefano Ermon, and Chelsea Finn. 2024. Direct preference optimization: Your language model is secretly a reward model. Advances in Neural Information Processing Systems, 36

    2024

  54. [54]

    N Reimers. 2019. Sentence-bert: Sentence embeddings using siamese bert-networks. arXiv preprint arXiv:1908.10084

    work page internal anchor Pith review arXiv 2019

  55. [55]

    Sahand Sabour, Siyang Liu, Zheyuan Zhang, June M Liu, Jinfeng Zhou, Alvionna S Sunaryo, Juanzi Li, Tatia MC Lee, Rada Mihalcea, and Minlie Huang. 2024. Emobench: Evaluating the emotional intelligence of large language models. CoRR

    2024

  56. [56]

    Sahand Sabour, Chujie Zheng, and Minlie Huang. 2022. Cem: Commonsense-aware empathetic response generation. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 36, pages 11229--11237

    2022

  57. [57]

    Maarten Sap, Hannah Rashkin, Derek Chen, Ronan LeBras, and Yejin Choi. 2019. Socialiqa: Commonsense reasoning about social interactions. arXiv preprint arXiv:1904.09728

    work page internal anchor Pith review arXiv 2019

  58. [58]

    Tim Schopf, Dennis N Schneider, and Florian Matthes. 2023. Efficient domain adaptation of sentence embeddings using adapters. arXiv preprint arXiv:2307.03104

    work page arXiv 2023

  59. [59]

    John Schulman, Filip Wolski, Prafulla Dhariwal, Alec Radford, and Oleg Klimov. 2017. Proximal policy optimization algorithms. arXiv preprint arXiv:1707.06347

    work page internal anchor Pith review arXiv 2017

  60. [60]

    Zhihong Shao, Yeyun Gong, Yelong Shen, Minlie Huang, Nan Duan, and Weizhu Chen. 2023. Synthetic prompting: Generating chain-of-thought demonstrations for large language models. In International Conference on Machine Learning, pages 30706--30775. PMLR

    2023

  61. [61]

    Jocelyn Shen. 2023. Modeling empathic similarity in personal narratives. Ph.D. thesis, Massachusetts Institute of Technology

    2023

  62. [62]

    Taiwei Shi, Kai Chen, and Jieyu Zhao. 2023. Safer-instruct: Aligning language models with automated preference data. arXiv preprint arXiv:2311.08685

    work page arXiv 2023

  63. [63]

    Feifan Song, Bowen Yu, Minghao Li, Haiyang Yu, Fei Huang, Yongbin Li, and Houfeng Wang. 2024. Preference ranking optimization for human alignment. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 18990--18998

    2024

  64. [64]

    Md Arafat Sultan, Jatin Ganhotra, and Ram \'o n Fernandez Astudillo. 2024. Structured chain-of-thought prompting for few-shot generation of content-grounded qa conversations. arXiv preprint arXiv:2402.11770

    work page arXiv 2024

  65. [65]

    Zhiqing Sun, Sheng Shen, Shengcao Cao, Haotian Liu, Chunyuan Li, Yikang Shen, Chuang Gan, Liang-Yan Gui, Yu-Xiong Wang, Yiming Yang, et al. 2023. Aligning large multimodal models with factually augmented rlhf. arXiv preprint arXiv:2309.14525

    work page arXiv 2023

  66. [66]

    Jinzhe Tan, Hannes Westermann, Nikhil Reddy Pottanigari, Jarom \' r S avelka, S \'e bastien Mee \`u s, Mia Godet, and Karim Benyekhlef. 2024. Robots in the middle: Evaluating llms in dispute resolution. In Legal Knowledge and Information Systems, pages 168--179. IOS Press

    2024

  67. [67]

    Hashimoto

    Rohan Taori, Ishaan Gulrajani, Tianyi Zhang, Yann Dubois, Xuechen Li, Carlos Guestrin, Percy Liang, and Tatsunori B. Hashimoto. 2023. Stanford alpaca: An instruction-following llama model. https://github.com/tatsu-lab/stanford_alpaca

    2023

  68. [68]

    Michael Henry Tessler, Michiel A Bakker, Daniel Jarrett, Hannah Sheahan, Martin J Chadwick, Raphael Koster, Georgina Evans, Lucy Campbell-Gillingham, Tantum Collins, David C Parkes, et al. 2024. Ai can help humans find common ground in democratic deliberation. Science, 386(6719):eadq2852

    2024

  69. [69]

    Philippe Van Parijs. 2012. What makes a good compromise? 1. Government and Opposition, 47(3):466--480

    2012

  70. [70]

    Haoxiang Wang, Yong Lin, Wei Xiong, Rui Yang, Shizhe Diao, Shuang Qiu, Han Zhao, and Tong Zhang. 2024 a . Arithmetic control of llms for diverse user preferences: Directional preference alignment with multi-objective rewards. arXiv preprint arXiv:2402.18571

    work page arXiv 2024

  71. [71]

    Liang Wang, Nan Yang, Xiaolong Huang, Binxing Jiao, Linjun Yang, Daxin Jiang, Rangan Majumder, and Furu Wei. 2022. Text embeddings by weakly-supervised contrastive pre-training. arXiv preprint arXiv:2212.03533

    work page internal anchor Pith review arXiv 2022

  72. [72]

    Xuena Wang, Xueting Li, Zi Yin, Yue Wu, and Jia Liu. 2023. Emotional intelligence of large language models. Journal of Pacific Rim Psychology, 17:18344909231213958

    2023

  73. [73]

    Zhichao Wang, Bin Bi, Shiva Kumar Pentyala, Kiran Ramnath, Sougata Chaudhuri, Shubham Mehrotra, Xiang-Bo Mao, Sitaram Asur, et al. 2024 b . A comprehensive survey of llm alignment techniques: Rlhf, rlaif, ppo, dpo and more. arXiv preprint arXiv:2407.16216

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  74. [74]

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

    2022

  75. [75]

    Anuradha Welivita, Yubo Xie, and Pearl Pu. 2021. A large-scale dataset for empathetic response generation. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 1251--1264

    2021

  76. [76]

    Daniel Wesego and Pedram Rooshenas. 2024. https://openreview.net/forum?id=JbuP6UV3Fk Score-based multimodal autoencoder . Transactions on Machine Learning Research

    2024

  77. [77]

    Ruoxi Xu, Hongyu Lin, Xianpei Han, Le Sun, and Yingfei Sun. 2024. Academically intelligent llms are not necessarily socially intelligent. arXiv preprint arXiv:2403.06591

    work page arXiv 2024

  78. [78]

    Yosuke Yamagishi and Yuta Nakamura. 2024. Utrad-nlp at\# smm4h 2024: Why llm-generated texts fail to improve text classification models. In Proceedings of The 9th Social Media Mining for Health Research and Applications (SMM4H 2024) Workshop and Shared Tasks, pages 42--47

    2024

  79. [79]

    Yueqin Yin, Zhendong Wang, Yi Gu, Hai Huang, Weizhu Chen, and Mingyuan Zhou. 2024. Relative preference optimization: Enhancing llm alignment through contrasting responses across identical and diverse prompts. arXiv preprint arXiv:2402.10958

    work page arXiv 2024

  80. [80]

    Hongyi Yuan, Zheng Yuan, Chuanqi Tan, Wei Wang, Songfang Huang, and Fei Huang. 2024. Rrhf: Rank responses to align language models with human feedback. Advances in Neural Information Processing Systems, 36

    2024

Showing first 80 references.