Pith. sign in

REVIEW 4 major objections 7 minor 1 cited by

Evaluating Multi-Turn Bargain Skills in LLM-Based Seller Agent

T0 review · 4 major / 7 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A turn-level evaluation framework shows LLM seller agents track buyer intent in bargaining only about half the time.

desk verdict A large, well-structured synthetic benchmark for turn-level intent recognition in bargaining, but the LLM-attached gold labels are unvalidated, so the F1 numbers may measure agreement with the generator rather than bargaining skill. read the letter →

arxiv 2509.06341 v1 pith:W76QLLMO submitted 2025-09-08 cs.AI

Issue Yishu Wang , Kakam Chong , Xiaofeng Wang , Xu Yan , DeXin Kong , Chen Ju , Ming Chen , Shuai Xiao
show 2 more authors
Shuguang Han jufeng chen
This is my paper · ORCID
classification cs.AI
keywords multi-turnbargainingselleragentintenttrackingtheoryofmindLLMevaluatione-commercedialogueintent-action-toolhierarchybenchmark
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 paper argues that a seller agent's bargaining ability should be measured by whether it can extract and track the buyer's intentions across turns, not just by whether a deal is reached. To make that measurable, it introduces BargainBench, with 3,014 scripted tasks built from 9,892 real product listings across 622 categories and turn-level ground-truth intent labels produced by an automated pipeline. The point of the benchmark is to test the intermediate reasoning that separates genuine negotiation from surface imitation. Its headline finding is that current LLM seller agents are far from reliable at this core skill: the best system reaches 56.7% intent precision, and the best F1 balance is 55.0% at Turn-3.

What carries the argument

The carrying mechanism is the intent–action–tool hierarchy, in which an intent is a coarse goal, an action is a mid-level negotiation move, and a tool is an atomic, directly checkable operation such as querying a price. The Intent Factory distills this hierarchy from 10,000 authentic marketplace dialogues using an extractor–verifier–maintainer pipeline that keeps coverage above 95% while removing duplicates. The Problem Weaver samples a real product and an ordered intent sequence and prompts an LLM to write a natural buyer question for each intent, producing scripted dialogues with ground-truth labels. The Evaluation Center then feeds each dialogue, product information, and a 20-option intent choice space to the target model and scores its per-turn predictions with precision, recall, F1, and failure rate, turning open-ended negotiation into a closed-set intent-tracking task.

What would settle it

Take a random sample of benchmark tasks, have human annotators write buyer questions that express the same ground-truth intents without seeing the LLM-generated scripts, and rerun the same graders; if model F1 drops substantially or the system ranking changes, the benchmark's measure depends on the generating model's phrasing rather than on intent tracking.

Watch

Extended reading notes

Core claim

The paper's central claim is that multi-turn bargaining ability can be decomposed into an intent–action–tool hierarchy and measured turn by turn, and that doing so reveals a clear gap in current models. Evaluated at the intent level, GPT-5-chat combines the highest precision (56.7%) with near-zero failure, while Qwen2.5-72B-Instruct achieves the best F1 (55.0% at Turn-3) through stronger recall. The discovery also includes the pattern that precision separates strong from weak systems more than recall does, and that longer dialogues mainly increase inconsistency rather than missed coverage. The authors read these results as evidence that even advanced LLMs do not reliably track what a buyer wants over the course of a negotiation, which is exactly the skill a trustworthy seller agent needs.

Load-bearing premise

All ground-truth labels are assigned by the same automated LLM pipeline that writes the buyer questions, so if those synthetic questions do not capture how real buyers phrase their intents, the benchmark scores style agreement rather than bargaining skill.

Editorial extensions

If this is right

  • Structured intents are already recognizable: authenticity checks, terminology explanations, and policy lookups score 83–87%, while rare or ambiguous intents such as promoting a logistics service and business cooperation fall to 4–17%.
  • The best current models still leave most bargaining intent unresolved: GPT-5-chat tops precision at 56.7%, Qwen2.5-72B-Instruct tops F1 at 55.0%, and a leading open-weight model collapses with failure rates above 50%.
  • Because precision, not recall, separates strong from weak models, interventions that reduce confident wrong-intent selections should improve measured bargaining skill more than expanding what counts as a correct intent.
  • Longer dialogues help up to a point and then hurt: several models improve from Turn-2 to Turn-3 but decline at Turn-4+, suggesting that inconsistency, not missing intents, is the binding constraint.
  • The framework's loop of intent extraction, scenario synthesis, and turn-level grading can be re-targeted to other goal-oriented domains such as diplomacy, persuasion, or multi-party coordination.

Reading between the lines

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

  • An untested implication is that the benchmark measures agreement with the generating model's phrasing style rather than bargaining skill; replacing the synthetic buyer questions with human-written paraphrases of the same intents would settle this.
  • Because mismatched-but-valid intent predictions are common, the harder failure mode is choosing among plausible candidates, so an error analysis separating mismatched from invalid predictions could guide training more directly than the aggregate F1.
  • The same evaluation could be run with open-ended API-call output instead of multiple-choice selection, testing whether the intent-action-tool hierarchy grounds into executable actions rather than just label matching.
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 / 7 minor

Summary. The paper introduces BargainBench, a multi-turn evaluation framework for LLM-based seller agents in second-hand marketplace bargaining. The framework has three components: an Intent Factory that distills a three-level intent-action-tool hierarchy from 10k marketplace dialogues using qwen-plus-latest; a Problem Weaver that samples product-intent sequences and prompts an LLM to generate scripted buyer questions with gold intent labels attached by construction; and an Evaluation Center that scores seller models on turn-level intent prediction using precision, recall, F1, and failure rate. The authors evaluate nine LLMs and report that GPT-5-chat achieves the highest precision with zero failure rate, Qwen2.5-72B-Instruct achieves the best F1 at Turn-3 (55.0%), and DeepSeek-V3-671B performs worst. The central claim is that the benchmark measures bargaining ability by testing whether an agent can extract and track buyer intents.

Significance. If the ground-truth intent labels are valid, the benchmark is a useful step beyond outcome-only negotiation metrics: it is large (9,892 products, 3,014 tasks, 622 categories claimed), openly specifies a hierarchical intent space, and provides turn-level diagnostics. The reported results also make a falsifiable point that current frontier LLMs achieve only around 50% F1 at this task, which would be a meaningful finding. The framework is reproducible in principle because the pipeline is described in detail. However, the validity of every metric and ranking rests on the unvalidated, LLM-generated ground-truth labels, and the paper currently provides no human annotation, no error bars, and no baselines, so the significance can only be assessed conditional on those gaps being closed.

major comments (4)
  1. [Appendix B, Prompt 4 and 'Annotation' step] The gold intent is attached by construction rather than verified. The prompt's own worked example illustrates the problem: the instructed ground_truth_action is an ordered list [API_CheckHeightFit, API_QueryShippingPolicy, API_CalculateOfferPrice], and the instruction says 'Mention only the first API in the Ground Truth list,' yet the sample buyer_question ('My daughter is 135 cm—will the size 140 be too big for her? Could you do 50 yuan with free shipping?') simultaneously expresses all three intents. If generated questions routinely encode multiple intents while the gold label records only one, the evaluation penalizes models that correctly infer the additional intents, so the reported F1 values and model rankings are not trustworthy as measures of bargaining ability. The manuscript needs a human-annotated validation sample with agreement statistics, an explicit decision procedure for multi-intent turns, and an analysis of label noise.
  2. [Section 4, Task Formulation] The input description states that the intent choice space is 'a set of 20 candidate options randomly sampled from the complete intent space,' but it does not state that the gold intent is always included. If the gold is not guaranteed to be among the candidates, precision and recall are not well-defined as stated, and the comparison across models is compromised. Please specify the sampling protocol (e.g., gold plus 19 distractors) and, if the gold is sometimes absent, report performance conditioned on gold presence.
  3. [Section 5, Table 2] All reported numbers come from a single evaluation pass with no error bars, bootstrap intervals, or statistical significance tests. Several ranking statements rest on small differences: for example, Turn-3 precision is 56.73 for GPT-5-chat versus 53.77 for Qwen2.5-72B-Instruct, and Turn-3 F1 is 55.02 versus 52.24. Without run-to-run variation or paired tests, the paper cannot support the claim that one system is 'the strongest' or that Qwen is 'competitive on F1.' Add repeated evaluations or confidence intervals before drawing comparative conclusions.
  4. [Appendix D and Table 2] The intent space is extracted and refined by qwen-plus-latest, and several of the evaluated systems are Qwen models (qwen2.5-72b-instruct, qwen3-14b, qwen3-32b). Because the task is to predict intents from a choice space derived by the same model family, higher Qwen scores could reflect familiarity with the generator's ontology and phrasing rather than bargaining skill. The authors should address this circularity concern by validating labels with human annotators (as in the first major comment) or by showing that a non-Qwen extraction produces equivalent model rankings.
minor comments (7)
  1. [Abstract and Table 1] The abstract reports '622 categories,' but Table 1 lists 85 level-1, 700 level-2, 1,336 level-3, and 1,611 level-4 unique categories; please reconcile the number or specify which hierarchy level is meant.
  2. [Table 2 header] The header contains the typo 'Failurs' instead of 'Failure'; please correct it.
  3. [Prompt 4] The output-format instruction contains the typo 'dineer' (likely 'dinner'); please correct it.
  4. [Section 4, Metrics] The definitions of CI, MMI, MI, and II do not specify how multiple predictions per turn are handled in the denominators; please state explicitly whether a model outputs one intent per turn or an ordered list, and how extra or duplicate predictions are counted.
  5. [Abstract and Section 1] The phrase 'grounded in Theory of Mind' overstates the evaluation, which measures intent identification rather than belief or desire inference; please either justify the ToM framing or soften the terminology.
  6. [Section 5 and Table 2] No random or majority-class baseline is reported; adding one would help interpret the 45-55% F1 range, since random selection among 20 options would give roughly 5% precision.
  7. [Appendix A, Coverage formula] The coverage metric is defined as Coverage = M/G, but the text does not explain how the ground-truth intents G in the held-out dialogue set are themselves determined, i.e., by whom or by what procedure; please clarify.

Circularity Check

0 steps flagged · score 2.0 of 10

No material circularity; the benchmark's ground-truth construction is a validity caveat, not a derivation-level loop.

full rationale

The paper's claimed derivation chain is: real dialogues supply the intent space (Section 3.1, Appendix D); the Problem Weaver samples a product-intent sequence, prompts an LLM to write a buyer question that 'naturally triggers' that sequence, and attaches the sampled sequence as the gold label (Section 3.2, Appendix B); the Evaluation Center then scores target LLMs by exact-match against those labels (Section 3.3). No fitted parameter is renamed a prediction, no uniqueness theorem is imported from the authors' prior work, and no equation reduces a reported result to an input by construction. The F1 and precision values in Table 2 are genuine measurements under the paper's operationalization. The only circularity-adjacent items are minor self-citations: 'integrating prior context Dexin and Xu [2025]' (FishBargain, authored by two of the present co-authors) and 'debt collection frameworks Wang et al. [2025]' (co-author Xiaofeng Wang). These are background references and are not load-bearing for the benchmark construction or the model rankings. The more serious issue is construct validity rather than circularity: Appendix B Step 4 attaches gold labels by construction, and Appendix D identifies qwen-plus-latest as the extractor, so the labels are never independently checked against human annotation. Prompt 4's own illustrative output ('My daughter is 135 cm—will the size 140 be too big for her? Could you do 50 yuan with free shipping?') also appears to trigger more than the first listed API, suggesting label ambiguity. These are unvalidated-assumption and label-noise concerns; they do not make the evaluation's outputs equal to its inputs in a derivation-level sense.

Assumptions & free parameters 1 free parameters · 4 assumptions · 0 invented entities

The central claim rests on the assumptions that the intent-action-tool tree faithfully represents real buyer behavior, that the LLM-generated buyer questions express the sampled intents in a realistic way, and that turn-level intent accuracy is a valid proxy for bargaining ability. The only hand-picked numerical parameter that directly shapes task difficulty is the choice space size of 20. No new physical or formal entities are introduced; the benchmark components are measurement artifacts.

free parameters (1)
  • choice_space_size = 20
    The number of candidate intents per turn is fixed at 20 by hand; it affects task difficulty and the expected chance level, but no justification or sensitivity analysis is given.
assumptions (4)
  • domain assumption The three-level intent-action-tool hierarchy is a faithful decomposition of buyer intents in bargaining dialogues.
    Invoked throughout Section 3.1; no human agreement study confirms that the taxonomy matches how buyers and sellers actually categorize intents.
  • ad hoc to paper The LLM-generated buyer questions in Problem Weaver naturally express the sampled intents.
    Stated as the generation rule in Section 3.2; the labels are assigned by construction rather than by independent annotation, so validity depends on the generator's skill.
  • domain assumption The extracted intent space covers real buyer intents with coverage above 95%.
    Appendix A measures coverage against LLM-extracted ground-truth intents from held-out dialogues, not against human annotations.
  • domain assumption Turn-level intent accuracy is a meaningful proxy for overall bargaining ability.
    The paper motivates this in Sections 1 and 3, but no correlation with deal outcomes or human judgments is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Evaluating Multi-Turn Bargain Skills in LLM-Based Seller Agent." pith.science (2026). https://pith.science/paper/W76QLLMO

@misc{pith2026250906341,
  author       = {Pith},
  title        = {Pith review of: Evaluating Multi-Turn Bargain Skills in LLM-Based Seller Agent},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/W76QLLMO}},
  note         = {Machine review of arXiv:2509.06341}
}
read the original abstract

In online second-hand marketplaces, multi-turn bargaining is a crucial part of seller-buyer interactions. Large Language Models (LLMs) can act as seller agents, negotiating with buyers on behalf of sellers under given business constraints. A critical ability for such agents is to track and accurately interpret cumulative buyer intents across long negotiations, which directly impacts bargaining effectiveness. We introduce a multi-turn evaluation framework for measuring the bargaining ability of seller agents in e-commerce dialogues. The framework tests whether an agent can extract and track buyer intents. Our contributions are: (1) a large-scale e-commerce bargaining benchmark spanning 622 categories, 9,892 products, and 3,014 tasks; (2) a turn-level evaluation framework grounded in Theory of Mind (ToM) with annotated buyer intents, moving beyond outcome-only metrics; and (3) an automated pipeline that extracts reliable intent from massive dialogue data.

Figures

Figures reproduced from arXiv: 2509.06341 by the authors.

Figure 1
Figure 1. Left: Human sellers se￾lect items and delegate them to the AI seller agent. Right: Human buy￾ers interact with the AI agent to ask questions and negotiate prices Bargaining is a fundamental social intelligence skill, widely applied in e-commerce, diplomacy, and labor negotiationsHe et al. [2018]. It requires the interpretation of scenario-specific information, reasoning about counterpart goals and constraints Davids… view at source ↗
Figure 2
Figure 2. BargainBench framework: Intent Factory extracts an intent space, Problem Weaver generates scripted dialogues, and Evaluation Center scores LLM performance. Our work makes three main contributions. First, we present a bargaining benchmark on e-commerce which is substantially larger and more challenging than prior datasets. It cover 622 categories, 9,892 product listings, and 3,014 tasks. Second, we introduce a turn-l… view at source ↗
Figure 3
Figure 3. Tools vs. Intents distributions. 4 [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: left: Effect of progressively adding modules refine intent number; right: Convergence [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Intent 66: finalized version of hierarchical intent space. [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Pie chart of listed item Level 1 categories. i [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: Pie chart of intent space. 8 [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: An illustration of evaluation sample [PITH_FULL_IMAGE:figures/full_fig_p009_8.png]
Figure 9
Figure 9. Figure 9: Candidate Intents F Extended Related Work Here we include the extended discussion of related work. Multi-Turn Interaction and Negotiation Benchmarks. Early datasets such as DealOrNoDeal Lewis et al. [2017] and CraigslistBargain He et al. [2018] established text-based b…

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Can LLM Agents Price Competitively? A Dynamic Multi-Attribute Auction Benchmark for Agentic Commerce

    cs.AI 2026-07 conditional novelty 7.0 of 10

    LLM merchant agents in a new dynamic auction benchmark capture at most 32% of hindsight-optimal profit; profit tracks margin per win more than win rate, and fast pre-shock learners adapt poorly to preference shocks.

Reference graph

Works this paper leans on

13 extracted references · 1 canonical work pages · cited by 1 Pith paper

  1. [1]

    MultiWOZ - A Large - Scale Multi - Domain Wizard -of- Oz Dataset for Task - Oriented Dialogue Modelling

    Paweł Budzianowski, Tsung-Hsien Wen, Bo-Hsiang Tseng, Iñigo Casanueva, Stefan Ultes, Osman Ramadan, and Milica Gašić. MultiWOZ - A Large - Scale Multi - Domain Wizard -of- Oz Dataset for Task - Oriented Dialogue Modelling . In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing , pages 5016--5026, Brussels, Belgium, 2018...

    doi:10.18653/v1/d18-1547 2018

  2. [2]

    NegotiationToM : A Benchmark for Stress -testing Machine Theory of Mind on Negotiation Surrounding

    Chunkit Chan, Cheng Jiayang, Yauwai Yim, Zheye Deng, Wei Fan, Haoran Li, Xin Liu, Hongming Zhang, Weiqi Wang, and Yangqiu Song. NegotiationToM : A Benchmark for Stress -testing Machine Theory of Mind on Negotiation Surrounding . In Findings of the Association for Computational Linguistics : EMNLP 2024 , pages 4211--4241, Miami, Florida, USA, 2024. Associa...

    doi:10.18653/v1/2024.findings-emnlp.244 2024

  3. [3]

    ACEBench : Who Wins the Match Point in Tool Usage ?, July 2025

    Chen Chen, Xinlong Hao, Weiwen Liu, Xu Huang, Xingshan Zeng, Shuai Yu, Dexun Li, Shuai Wang, Weinan Gan, Yuefeng Huang, Wulong Liu, Xinzhi Wang, Defu Lian, Baoqun Yin, Yasheng Wang, and Wu Liu. ACEBench : Who Wins the Match Point in Tool Usage ?, July 2025. URL http://arxiv.org/abs/2501.12851. arXiv:2501.12851 [cs]

    arXiv 2025

  4. [4]

    Davidson, Veniamin Veselovsky, Martin Josifoski, Maxime Peyrard, Antoine Bosselut, Michal Kosinski, and Robert West

    Tim R. Davidson, Veniamin Veselovsky, Martin Josifoski, Maxime Peyrard, Antoine Bosselut, Michal Kosinski, and Robert West. Evaluating Language Model Agency through Negotiations . arXiv, March 2024. doi:10.48550/arXiv.2401.04536. URL http://arxiv.org/abs/2401.04536. arXiv:2401.04536 [cs]

    work page doi:10.48550/arxiv.2401.04536 2024

  5. [5]

    FishBargain : An LLM - Empowered Bargaining Agent for Online Fleamarket Platform Sellers

    Kong Dexin and Yan Xu. FishBargain : An LLM - Empowered Bargaining Agent for Online Fleamarket Platform Sellers . Erscheinungsort nicht ermittelbar, 2025. Association for Computing Machinery. ISBN 979-8-4007-1331-6. doi:10.1145/3701716

    doi:10.1145/3701716 2025

  6. [6]

    Evaluating LLM -based Agents for Multi - Turn Conversations : A Survey , March 2025

    Shengyue Guan, Haoyi Xiong, Jindong Wang, Jiang Bian, Bin Zhu, and Jian-guang Lou. Evaluating LLM -based Agents for Multi - Turn Conversations : A Survey , March 2025. URL http://arxiv.org/abs/2503.22458. arXiv:2503.22458 [cs]

    arXiv 2025

  7. [7]

    Decoupling Strategy and Generation in Negotiation Dialogues

    He He, Derek Chen, Anusha Balakrishnan, and Percy Liang. Decoupling Strategy and Generation in Negotiation Dialogues . In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing , pages 2333--2343, Brussels, Belgium, 2018. Association for Computational Linguistics. doi:10.18653/v1/D18-1256. URL http://aclweb.org/anthology/D18-1256

    doi:10.18653/v1/d18-1256 2018

  8. [8]

    Deal or No Deal ? End -to- End Learning of Negotiation Dialogues

    Mike Lewis, Denis Yarats, Yann Dauphin, Devi Parikh, and Dhruv Batra. Deal or No Deal ? End -to- End Learning of Negotiation Dialogues . In Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing , pages 2443--2453, Copenhagen, Denmark, 2017. Association for Computational Linguistics. doi:10.18653/v1/D17-1259. URL http://acl...

    doi:10.18653/v1/d17-1259 2017

Show all 13 references
  1. [9]

    ToolACE : Winning the Points of LLM Function Calling , July 2025

    Weiwen Liu, Xu Huang, Xingshan Zeng, Xinlong Hao, Shuai Yu, Dexun Li, Shuai Wang, Weinan Gan, Zhengying Liu, Yuanqing Yu, Zezhong Wang, Yuxian Wang, Wu Ning, Yutai Hou, Bin Wang, Chuhan Wu, Xinzhi Wang, Yong Liu, Yasheng Wang, Duyu Tang, Dandan Tu, Lifeng Shang, Xin Jiang, Rui...

    2025 arXiv

  2. [10]

    Debt Collection Negotiations with Large Language Models : An Evaluation System and Optimizing Decision Making with Multi - Agent , February 2025

    Xiaofeng Wang, Zhixin Zhang, Jinguang Zheng, Yiming Ai, and Rui Wang. Debt Collection Negotiations with Large Language Models : An Evaluation System and Optimizing Decision Making with Multi - Agent , February 2025. URL http://arxiv.org/abs/2502.18228. arXiv:2502.18228 [cs]

    2025 arXiv

  3. [11]

    Measuring Bargaining Abilities of LLMs : A Benchmark and A Buyer - Enhancement Method

    Tian Xia, Zhiwei He, Tong Ren, Yibo Miao, Zhuosheng Zhang, Yang Yang, and Rui Wang. Measuring Bargaining Abilities of LLMs : A Benchmark and A Buyer - Enhancement Method . arXiv, June 2024. doi:10.48550/arXiv.2402.15813. URL http://arxiv.org/abs/2402.15813. arXiv:2402.15813 [cs]

    2024 doi arXiv

  4. [12]

    \ τ\ -bench: A Benchmark for Tool - Agent - User Interaction in Real - World Domains

    Shunyu Yao, Noah Shinn, Pedram Razavi, and Karthik Narasimhan. \ τ\ -bench: A Benchmark for Tool - Agent - User Interaction in Real - World Domains . arXiv, June 2024. doi:10.48550/arXiv.2406.12045. URL http://arxiv.org/abs/2406.12045. arXiv:2406.12045 [cs]

    2024 doi arXiv

  5. [13]

    SOTOPIA : Interactive Evaluation for Social Intelligence in Language Agents , March 2024

    Xuhui Zhou, Hao Zhu, Leena Mathur, Ruohong Zhang, Haofei Yu, Zhengyang Qi, Louis-Philippe Morency, Yonatan Bisk, Daniel Fried, Graham Neubig, and Maarten Sap. SOTOPIA : Interactive Evaluation for Social Intelligence in Language Agents , March 2024. URL http://arxiv.org/abs/231...

    2024 arXiv

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.