Pith. sign in

REVIEW 3 major objections 4 minor 41 references

A Practical Approach for Building Production-Grade Conversational Agents with Workflow Graphs

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read A workflow graph splits a conversational agent into per-state prompts and tools, and response-masked fine-tuning lifts task accuracy 52% and format adherence 50% over a single-prompt baseline, beating GPT-4o.

desk verdict Solid industrial paper with a genuinely useful fine-tuning trick, but the headline 52%/50% improvements are not reproducible from Table 1 as stated. read the letter →

arxiv 2505.23006 v1 pith:LLU2MIGU submitted 2025-05-29 cs.CL cs.AI

Chiwan Park , Wonjun Jang , Daeryong Kim , Aelim Ahn , Kichang Yang , Woosung Hwang , Jihyeon Roh , Hyerin Park
show 3 more authors
Hyosun Wang Min Seok Kim Jihoon Kang
This is my paper
classification cs.CLcs.AI
keywords conversationalagentsworkflowgraphdirectedacyclice-commercefine-tuningresponsemaskingLLM-as-a-Judgeformatadherence
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

The paper tries to establish a recipe for building conversational agents that must be both flexible and strictly compliant, a tension that monolithic prompt-based LLMs resolve poorly. Its claim is that modeling the agent as a directed acyclic graph—where each node carries its own system prompt, tools, and business rules—and then fine-tuning the LLM on conversations while masking the loss of responses generated by other nodes, yields large, repeatable gains. In a private test set of 161 annotated e-commerce conversations, this approach improved task accuracy by up to 52% and format adherence by up to 50% over a single-prompt baseline across three model families, and pushed a 27B-32B internal model past GPT-4o on accuracy, format adherence, and response quality. A human preference battle test on the deployed service found the internal model preferred over the GPT-4o-based agent in safety, product recommendation, and messenger-related turns, while GPT-4o won regular chat. If the recipe works as reported, it gives production teams a practical path to controllable LLM agents without sacrificing general conversational ability.

What carries the argument

The load-bearing mechanism is the workflow graph, a directed acyclic graph whose nodes are either LLM calls with state-specific system prompts and custom history-manipulation routines, or external tool calls with input/output schemas; traversal of the graph replaces end-to-end generation, so each turn is governed by the short, focused instructions of its current state. The second mechanism is response-masked fine-tuning: the training sequence for a node pairs its system prompt with the full conversation, but the loss is computed only on responses that the node itself produced, so the model never learns to imitate outputs that obey a different node's constraints. The final piece is the data pipeline: a prototype agent built from the graph generates draft responses that human annotators correct, and constrained decoding forces LLM outputs that feed tool nodes to be syntactically valid JSON.

What would settle it

Have three independent human annotators re-score the 161 test conversations with the same accuracy, format, and quality criteria, and compute agreement with the o3-mini judge; if agreement is low (for example, Cohen's kappa below 0.5 on the binary metrics), the reported 52% and 50% improvements would be artifacts of the judge rather than measurable differences in response quality.

Watch

Extended reading notes

Core claim

The central claim is that the way an agent is structured, not just the size of its model, determines whether it can satisfy hard business constraints in production. The paper shows that replacing a single system prompt with a workflow graph—each LLM node having its own shorter prompt, few-shot examples, optional history-manipulation routine, and tool schemas—improves accuracy by up to 14% and format adherence by up to 45% relative to the monolithic baseline for the same model. Adding the proposed fine-tuning, in which the training loss is masked for any response that originated from a different graph node than the one being trained, brings further gains: the internal 27B-32B model reaches 0.890 task accuracy and 0.987 format adherence, versus 0.864 and 0.778 for the GPT-4o basic agent and 0.888 and 0.964 for the GPT-4o graph agent. The authors interpret this as evidence that graph-level decomposition plus node-aware training corrects both hallucinated tool calls and formatting violations that even state-of-the-art LLMs exhibit when given long, compound instructions.

Load-bearing premise

The headline gains assume that a private test set of 161 conversations, scored by an LLM judge (o3-mini) against annotator-corrected references, actually reflects what users of the deployed service experience, a concern the paper's own limitations section raises.

Editorial extensions

If this is right

  • Because the same workflow-graph architecture improved accuracy and format adherence for Qwen 2.5 32B, Gemma 3 27B, and the internal 27B-32B model, the recipe appears to generalize across open and proprietary model families rather than being specific to one checkpoint.
  • The fine-tuned graph agents let 27B-32B open-source models match or beat GPT-4o on this e-commerce workload, which implies that production teams can hit high compliance with smaller, cheaper, lower-latency models when the workflow and training are built around the graph structure.
  • The deployed service, live since December 2024 and covering over one million products, demonstrates that the approach survives real-world traffic and platform constraints, and the authors state the framework is adaptable to other domains with complex task execution.
  • The human battle test adds evidence that users prefer the graph-fine-tuned internal model over the GPT-4o-based agent in safety, recommendation, and messenger feature scenarios, with GPT-4o retaining an edge only in general chit-chat.

Reading between the lines

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

  • Because the gains concentrate in format adherence and tool-call correctness, the graph is acting as a form of structured prior that reduces the effective instruction-following burden on the LLM; a testable extension would be to measure per-node prompt lengths and latency to verify the paper's implicit claim that shorter prompts are the main driver of the improvement.
  • The response-masking scheme is a simple, model-agnostic fix for a problem that any multi-state agent framework faces during fine-tuning, so it could plausibly be dropped into other graph-based agent frameworks that route conversations through multiple LLM nodes.
  • The human-vs-LLM-judge discrepancy the authors note—fluency matters to users but not to the automated judge—suggests that production evaluations should pair LLM-as-a-Judge with periodic human sampling; quantifying the human-judge agreement would be a direct follow-up.
  • Since the prototype agent used GPT-4o to generate draft responses before human correction, the resulting training data may carry a stylistic residue of GPT-4o; one could test whether using a weaker prototype model changes the fine-tuned agent's quality, which would bound the data pipeline's dependence on the initial generator.
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 / 4 minor

Summary. This industry case study proposes a hybrid architecture for a production e-commerce conversational agent: a directed acyclic graph (DAG) workflow in which each LLM-calling node has its own system prompt, tools, history-manipulation routine, and constrained decoding, combined with response-masked fine-tuning that excludes assistant turns generated by other graph nodes. The authors compare Basic, Workflow Graph (WG), and Workflow Graph with Fine-Tuning (WG-FT) variants on Qwen 2.5 32B, Gemma 3 27B, an internal 27B-32B Korean-tuned model, and GPT-4o, using a private 161-conversation test set with LLM-as-a-judge and a coded format validator. They report that WG-FT improves accuracy by 52% and format adherence by 50% over baseline and outperforms GPT-4o, and they support this with a human battle test on the deployed KakaoTalk service.

Significance. If the results hold, the paper offers a practical, deployable pattern for reconciling conversational flexibility with strict business-rule compliance: decomposing a complex agent into state-specific prompts and tools, then fine-tuning with node-aware loss masking. The strengths are the real-world deployment (AI Shopping Mate), a human-preference battle test with identical external systems, and a machine-readable format validator that gives objective measurement of one key dimension. The paper also honestly acknowledges two central limitations: the human-dependent data collection process and the lack of a systematic human/LLM-judge correlation study. The main weakness is that the headline quantitative claims are not supported by a single internally consistent baseline comparison, and the small private test set without statistical inference leaves the marginal claims over GPT-4o unquantified.

major comments (3)
  1. [Abstract and §4.2, Table 1] The evaluation rests on 161 conversations (2,100 turns) with no confidence intervals, significance tests, or variance estimates. Several headline comparisons are very small: Internal WG-FT accuracy (0.890) exceeds GPT-4o WG (0.888) by 0.002, and the response-quality differences are around 0.02–0.06 on a 1–3 scale. With 161 conversations, these differences are within plausible sampling variability. Please report bootstrap intervals or significance tests, and clarify whether the accuracy metric is computed per turn or per conversation.
  2. [§4.1–4.2, Table 1] The evaluation rests on 161 conversations (2,100 turns) with no confidence intervals, significance tests, or variance estimates. Several headline comparisons are very small: Internal WG-FT accuracy (0.890) exceeds GPT-4o WG (0.888) by 0.002, and the response-quality differences are around 0.02–0.06 on a 1–3 scale. With 161 conversations, these differences are within plausible sampling variability. Please report bootstrap intervals or significance tests, and clarify whether the accuracy metric is computed per turn or per conversation.
  3. [§3.2 and §6] The test set is drawn from the same annotator-corrected prototype-agent pipeline used for training, as acknowledged in the Limitations. Because both the references and the training data share the prototype's graph-traversal outputs, the lab metrics may overstate production quality; the human battle test in §4.3 is the only evaluation on real-world traffic but covers only internal model vs GPT-4o without the Basic or WG-FT ablation. Please add an evaluation on a held-out set of real user interactions, or state this limitation in the main text together with the human-battle results.
minor comments (4)
  1. [§1 and Appendix C] There are typos: "Antrophic" should be "Anthropic," and in the evaluation prompt in Figure 4, "non-understable characters" should be "non-understandable characters."
  2. [Table 1] Some table entries appear to be missing separating spaces, e.g., "0.7900.890" and "0.9510.987"; these should be rendered as distinct values in separate columns.
  3. [§3.1] The sentence "After reaching final you can restart from initial" is incomplete and should be rephrased, for example, "After reaching the final node, the graph can restart from the initial node for multi-turn conversations."
  4. [§4.1 and Table 1] The metric names are inconsistent: Table 1 uses "Response Validity" while §4.1 calls the third dimension "response quality." Unify the terminology throughout.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper's empirical claims are not derived from their inputs by construction, though the headline 52%/50% improvements are inconsistently attributed to different baselines.

full rationale

This is a systems/case-study paper with no mathematical derivation or predictive equation whose output is equivalent to an input. The workflow-graph design and response-masked fine-tuning are evaluated empirically. The training and test conversations both originate from the annotator-corrected prototype-agent pipeline (Sections 3.2 and 4.1); that is a standard supervised setup, and nothing in the text makes the test-set references a function of the fitted model, so it is not circular by construction. The only load-bearing quantitative claim in the abstract ('52% improvement in task accuracy and 50% improvement in format adherence compared to the baseline') is not reproducible from Table 1 for a single baseline: 52% matches Qwen B to WG-FT accuracy, while 50% matches internal B to WG-FT format adherence. This is an internal-consistency or overclaim problem, not a circularity problem. There are no self-citations used as load-bearing support, no imported uniqueness theorems, and no ansatz smuggled in via citation. The Limitations section explicitly concedes that 'LLM-as-a-Judge with reference answers... may not fully reflect human preferences,' but the deployed human battle test (Section 4.3) provides independent external grounding for the central claim. Consistent same-model gains (e.g., internal WG-FT versus B) remain observable in Table 1, so the core comparison is not forced by the evaluation design. Verdict: no significant circularity (score 0).

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

No free parameters or invented entities are required: the central contribution is an engineering configuration rather than a fitted scientific model. The paper does rely on several domain assumptions about the validity of its internal evaluation and the sufficiency of annotator-corrected prototype data; these are listed as axioms.

assumptions (4)
  • domain assumption o3-mini as LLM-as-a-Judge returns valid accuracy and response-quality scores.
    Used for accuracy and response validity in Section 4.1 with prompts in Appendix C; no correlation with human judgements is reported, and the Limitations say such correlation is future work.
  • domain assumption Annotator-corrected prototype-agent conversations are a representative sample of production user traffic.
    Data collection in Section 3.2 relies on human annotators interacting with a GPT-4o prototype; Limitations admit demographic bias and human dependency.
  • domain assumption Per-node system prompts with modify_history and constrained decoding reliably enforce compliance without hurting flexibility.
    Core design premise in Section 3.1; the paper does not ablate prompt length, history masking, or constrained decoding separately.
  • domain assumption Loss masking over node-specific responses prevents conflicting supervision and improves system-prompt fidelity.
    Mechanism described in Section 3.3; no controlled comparison of masked versus unmasked fine-tuning is presented.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Practical Approach for Building Production-Grade Conversational Agents with Workflow Graphs." pith.science (2026). https://pith.science/paper/LLU2MIGU

@misc{pith2026250523006,
  author       = {Pith},
  title        = {Pith review of: A Practical Approach for Building Production-Grade Conversational Agents with Workflow Graphs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LLU2MIGU}},
  note         = {Machine review of arXiv:2505.23006}
}
read the original abstract

The advancement of Large Language Models (LLMs) has led to significant improvements in various service domains, including search, recommendation, and chatbot applications. However, applying state-of-the-art (SOTA) research to industrial settings presents challenges, as it requires maintaining flexible conversational abilities while also strictly complying with service-specific constraints. This can be seen as two conflicting requirements due to the probabilistic nature of LLMs. In this paper, we propose our approach to addressing this challenge and detail the strategies we employed to overcome their inherent limitations in real-world applications. We conduct a practical case study of a conversational agent designed for the e-commerce domain, detailing our implementation workflow and optimizations. Our findings provide insights into bridging the gap between academic research and real-world application, introducing a framework for developing scalable, controllable, and reliable AI-driven agents.

Figures

Figures reproduced from arXiv: 2505.23006 by the authors.

Figure 1
Figure 1. A mobile messenger conversation between a user and our e-commerce agent. The first two turns require external tool calls to respond without hallucina￾tion. There are also output format constraints to make the responses readable in a mobile environment, such as emoji bullets. a product, and purchase the product. Furthermore, using social relationships in the messenger plat￾form, the user can check the birthdays of hi… view at source ↗
Figure 2
Figure 2. An example workflow graph. Each LLM calling node (green colored) has its system prompt and a custom [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Evaluation Prompt for Task Accuracy. You are requested to evaluate the linguistic quality of the generated response. You are given the following information as follows: - <history>: The chat history between the user, the model and the tool response. - <message>: the message that was sent by the user, the model or the tool. The sender of the message is given as ` role` attribute. - <response>: The response generated … view at source ↗
Figures from the paper (1 more)
Figure 5
Figure 5. Figure 5: Example use cases on AI Shopping Mate [PITH_FULL_IMAGE:figures/full_fig_p012_5.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

41 extracted references · 19 canonical work pages

  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]

    Antrophic. 2024. https://www-cdn.anthropic.com/de8ba9b01c9ab7cbabf5c33b80b7bbc618857627/Model_Card_Claude_3.pdf The claude 3 model family: Opus, sonnet, haiku . Technical report, Antrophic

    work page 2024

  4. [4]

    Guangyao Chen, Siwei Dong, Yu Shu, Ge Zhang, Jaward Sesay, B \" o rje Karlsson, Jie Fu, and Yemin Shi. 2024. https://www.ijcai.org/proceedings/2024/3 Autoagents: A framework for automatic agent generation . In IJCAI 2024 , pages 22--30

    work page 2024

  5. [5]

    Jordan, Joseph E

    Wei - Lin Chiang, Lianmin Zheng, Ying Sheng, Anastasios Nikolas Angelopoulos, Tianle Li, Dacheng Li, Banghua Zhu, Hao Zhang, Michael I. Jordan, Joseph E. Gonzalez, and Ion Stoica. 2024. https://openreview.net/forum?id=3MW8GKNyzI Chatbot arena: An open platform for evaluating llms by human preference . In Forty-first International Conference on Machine Lea...

    work page 2024

  6. [6]

    Anthony Costarelli, Mat Allen, Roman Hauksson, Grace Sodunke, Suhas Hariharan, Carlson Cheng, Wenjie Li, and Arjun Yadav. 2024. https://doi.org/10.48550/ARXIV.2406.06613 Gamebench: Evaluating strategic reasoning abilities of LLM agents . CoRR, abs/2406.06613

    arXiv doi:10.48550/arxiv.2406.06613 2024

  7. [7]

    DeepSeek - AI. 2024. https://doi.org/10.48550/ARXIV.2412.19437 Deepseek-v3 technical report . CoRR, abs/2412.19437

    arXiv doi:10.48550/arxiv.2412.19437 2024

  8. [8]

    Ruan, Yaxing Cai, Ruihang Lai, Ziyi Xu, Yilong Zhao, and Tianqi Chen

    Yixin Dong, Charlie F. Ruan, Yaxing Cai, Ruihang Lai, Ziyi Xu, Yilong Zhao, and Tianqi Chen. 2024. https://doi.org/10.48550/ARXIV.2411.15100 Xgrammar: Flexible and efficient structured generation engine for large language models . CoRR, abs/2411.15100

    arXiv doi:10.48550/arxiv.2411.15100 2024

Show all 41 references
  1. [9]

    Gemma Team . 2025. https://storage.googleapis.com/deepmind-media/gemma/Gemma3Report.pdf Gemma 3 technical report . Technical report, Google Deepmind

    2025

  2. [10]

    Charles Goddard, Shamane Siriwardhana, Malikeh Ehghaghi, Luke Meyers, Vladimir Karpukhin, Brian Benedict, Mark McQuade, and Jacob Solawetz. 2024. https://doi.org/10.18653/v1/2024.emnlp-industry.36 Arcee ' s M erge K it: A toolkit for merging large language models . In Proceedi...

    2024 doi

  3. [11]

    Juraj Gottweis, Wei-Hung Weng, Alexander Daryin, Tao Tu, Anil Palepu, Petar Sirkovic, Artiom Myaskovsky, Felix Weissenberger, Keran Rong, Ryutaro Tanno, Khaled Saab, Dan Popovici, Jacob Blum, Fan Zhang, Katherine Chou, Avinatan Hassidim, Burak Gokturk, Amin Vahdat, Pushmeet Ko...

    2025 arXiv

  4. [12]

    Shanshan Han, Qifan Zhang, Yuhang Yao, Weizhao Jin, Zhaozhuo Xu, and Chaoyang He. 2024. https://doi.org/10.48550/ARXIV.2402.03578 LLM multi-agent systems: Challenges and open problems . CoRR, abs/2402.03578

    2024 doi arXiv

  5. [13]

    Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen - Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen

    Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen - Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. 2022. https://openreview.net/forum?id=nZeVKeeFYf9 Lora: Low-rank adaptation of large language models . In The Tenth International Conference on Learning Representat...

    2022

  6. [14]

    Wenyue Hua, Xianjun Yang, Mingyu Jin, Zelong Li, Wei Cheng, Ruixiang Tang, and Yongfeng Zhang. 2024. https://doi.org/10.18653/v1/2024.findings-emnlp.585 T rust A gent: Towards safe and trustworthy LLM -based agents . In Findings of the Association for Computational Linguistics...

    2024 doi

  7. [15]

    Albert Q. Jiang, Alexandre Sablayrolles, Antoine Roux, Arthur Mensch, Blanche Savary, Chris Bamford, Devendra Singh Chaplot, Diego de Las Casas, Emma Bou Hanna, Florian Bressand, Gianna Lengyel, Guillaume Bour, Guillaume Lample, L \' e lio Renard Lavaud, Lucile Saulnier, Marie...

    2024 doi arXiv

  8. [16]

    Ahmet Baki Kocaballi, Emre Sezgin, Leigh Clark, John M Carroll, Yungui Huang, Jina Huh-Yoo, Junhan Kim, Rafal Kocielnik, Yi-Chieh Lee, Lena Mamykina, Elliot G Mitchell, Robert J Moore, Prasanth Murali, Elizabeth D Mynatt, Sun Young Park, Alessandro Pasta, Deborah Richards, Luc...

    2022 doi

  9. [17]

    Gonzalez, Hao Zhang, and Ion Stoica

    Woosuk Kwon, Zhuohan Li, Siyuan Zhuang, Ying Sheng, Lianmin Zheng, Cody Hao Yu, Joseph E. Gonzalez, Hao Zhang, and Ion Stoica. 2023. Efficient memory management for large language model serving with pagedattention. In Proceedings of the ACM SIGOPS 29th Symposium on Operating S...

    2023

  10. [18]

    Staffan Larsson and Alexander Berman. 2016. Domain-specific and general syntax and semantics in the talkamatic. In Empirical Issues in Syntax and Semantics 11, pages 91--110, Paris. CSSP

    2016

  11. [19]

    Mosh Levy, Alon Jacoby, and Yoav Goldberg. 2024. https://doi.org/10.18653/V1/2024.ACL-LONG.818 Same task, more tokens: the impact of input length on the reasoning performance of large language models . In Proceedings of the 62nd Annual Meeting of the Association for Computatio...

    2024 doi

  12. [20]

    Xiao Liu, Hao Yu, Hanchen Zhang, Yifan Xu, Xuanyu Lei, Hanyu Lai, Yu Gu, Hangliang Ding, Kaiwen Men, Kejuan Yang, Shudan Zhang, Xiang Deng, Aohan Zeng, Zhengxiao Du, Chenhui Zhang, Sheng Shen, Tianjun Zhang, Yu Su, Huan Sun, Minlie Huang, Yuxiao Dong, and Jie Tang. 2024. https...

    2024

  13. [21]

    Chang Ma, Junlei Zhang, Zhihao Zhu, Cheng Yang, Yujiu Yang, Yaohui Jin, Zhenzhong Lan, Lingpeng Kong, and Junxian He. 2024. http://papers.nips.cc/paper\_files/paper/2024/hash/877b40688e330a0e2a3fc24084208dfa-Abstract-Datasets\_and\_Benchmarks\_Track.html Agentboard: An analyti...

    2024

  14. [22]

    OpenAI. 2022. https://chat.openai.com Chatgpt

    2022

  15. [23]

    OpenAI. 2023. https://doi.org/10.48550/ARXIV.2303.08774 GPT-4 technical report . CoRR, abs/2303.08774

    2023 doi arXiv

  16. [24]

    Shuofei Qiao, Runnan Fang, Zhisong Qiu, Xiaobin Wang, Ningyu Zhang, Yong Jiang, Pengjun Xie, Fei Huang, and Huajun Chen. 2025. https://doi.org/10.48550/ARXIV.2410.07869 Benchmarking agentic workflow generation . In The Thirteenth International Conference on Learning Representa...

    2025 doi arXiv

  17. [25]

    Yujia Qin, Shihao Liang, Yining Ye, Kunlun Zhu, Lan Yan, Yaxi Lu, Yankai Lin, Xin Cong, Xiangru Tang, Bill Qian, Sihan Zhao, Lauren Hong, Runchu Tian, Ruobing Xie, Jie Zhou, Mark Gerstein, Dahai Li, Zhiyuan Liu, and Maosong Sun. 2024. https://openreview.net/forum?id=dHng2O0Jjr...

    2024

  18. [26]

    Rasa Technologies . 2019. Rasa: Open source conversational ai. https://rasa.com. Accessed: 2025-05-15

    2019

  19. [27]

    Calvo, and C \' e line Mougenot

    Malak Sadek, Rafael A. Calvo, and C \' e line Mougenot. 2023. https://doi.org/10.1145/3571884.3597143 Trends, challenges and processes in conversational agent design: Exploring practitioners' views through semi-structured interviews . In Proceedings of the 5th International Co...

    2023

  20. [28]

    Timo Schick, Jane Dwivedi - Yu, Roberto Dess \` , Roberta Raileanu, Maria Lomeli, Eric Hambro, Luke Zettlemoyer, Nicola Cancedda, and Thomas Scialom. 2023. http://papers.nips.cc/paper\_files/paper/2023/hash/d842425e4bf79ba039352da0f658a906-Abstract-Conference.html Toolformer: ...

    2023

  21. [29]

    Anubhav Shrimal, Stanley Kanagaraj, Kriti Biswas, Swarnalatha Raghuraman, Anish Nediyanchath, Yi Zhang, and Promod Yenigalla. 2024. https://aclanthology.org/2024.emnlp-industry.102 MARCO: multi-agent real-time chat orchestration . In Proceedings of the 2024 Conference on Empir...

    2024

  22. [30]

    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, Aur \' e lien Rodriguez, Armand Joulin, Edouard Grave, and Guillaume Lample. 2023. https://doi.org/10.485...

    2023 doi arXiv

  23. [31]

    Yancheng Wang, Ziyan Jiang, Zheng Chen, Fan Yang, Yingxue Zhou, Eunah Cho, Xing Fan, Yanbin Lu, Xiaojiang Huang, and Yingzhen Yang. 2024. https://doi.org/10.18653/V1/2024.FINDINGS-NAACL.271 Recmind: Large language model powered agent for recommendation . In Findings of the Ass...

    2024 doi

  24. [32]

    Chi, Quoc V

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Brian Ichter, Fei Xia, Ed H. Chi, Quoc V. Le, and Denny Zhou. 2022. http://papers.nips.cc/paper\_files/paper/2022/hash/9d5609613524ecf4f15af0f7b31abca4-Abstract-Conference.html Chain-of-thought prompting elicits reasoning...

    2022

  25. [33]

    Willard and R \' e mi Louf

    Brandon T. Willard and R \' e mi Louf. 2023. https://doi.org/10.48550/ARXIV.2307.09702 Efficient guided generation for large language models . CoRR, abs/2307.09702

    2023 doi arXiv

  26. [34]

    Jian Xie, Kai Zhang, Jiangjie Chen, Tinghui Zhu, Renze Lou, Yuandong Tian, Yanghua Xiao, and Yu Su. 2024. https://openreview.net/forum?id=l5XQzNkAOe Travelplanner: A benchmark for real-world planning with language agents . In Forty-first International Conference on Machine Lea...

    2024

  27. [35]

    An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, Huan Lin, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jiaxi Yang, Jingren Zhou, Junyang Lin, Kai Dang, Keming Lu, Keqin Bao, Kexin Yang, Le Yu, Me...

    2024 doi arXiv

  28. [36]

    Shunyu Yao, Dian Yu, Jeffrey Zhao, Izhak Shafran, Tom Griffiths, Yuan Cao, and Karthik Narasimhan. 2023 a . http://papers.nips.cc/paper\_files/paper/2023/hash/271db9922b8d1f4dd7aaef84ed5ac703-Abstract-Conference.html Tree of thoughts: Deliberate problem solving with large lang...

    2023

  29. [37]

    Narasimhan, and Yuan Cao

    Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik R. Narasimhan, and Yuan Cao. 2023 b . https://openreview.net/forum?id=WE\_vluYUL-X React: Synergizing reasoning and acting in language models . In The Eleventh International Conference on Learning Representation...

    2023

  30. [38]

    Kechi Zhang, Jia Li, Ge Li, Xianjie Shi, and Zhi Jin. 2024 a . https://doi.org/10.18653/V1/2024.ACL-LONG.737 Codeagent: Enhancing code generation with tool-integrated agent systems for real-world repo-level coding challenges . In Proceedings of the 62nd Annual Meeting of the A...

    2024 doi

  31. [39]

    Yuxiang Zhang, Jing Chen, Junjie Wang, Yaxin Liu, Cheng Yang, Chufan Shi, Xinyu Zhu, Zihao Lin, Hanwen Wan, Yujiu Yang, Tetsuya Sakai, Tian Feng, and Hayato Yamana. 2024 b . https://aclanthology.org/2024.emnlp-main.637 Toolbehonest: A multi-level hallucination diagnostic bench...

    2024

  32. [40]

    Xing, Hao Zhang, Joseph E

    Lianmin Zheng, Wei - Lin Chiang, Ying Sheng, Siyuan Zhuang, Zhanghao Wu, Yonghao Zhuang, Zi Lin, Zhuohan Li, Dacheng Li, Eric P. Xing, Hao Zhang, Joseph E. Gonzalez, and Ion Stoica. 2023. http://papers.nips.cc/paper\_files/paper/2023/hash/91f18a1287b398d378ef22505bf41832-Abstr...

    2023

  33. [41]

    Gonzalez, Clark Barrett, and Ying Sheng

    Lianmin Zheng, Liangsheng Yin, Zhiqiang Xie, Chuyue Sun, Jeff Huang, Cody Hao Yu, Shiyi Cao, Christos Kozyrakis, Ion Stoica, Joseph E. Gonzalez, Clark Barrett, and Ying Sheng. 2024. https://arxiv.org/abs/2312.07104 Sglang: Efficient execution of structured language model progr...

    2024 arXiv

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.