Pith. sign in

REVIEW 3 major objections 8 minor 4 cited by

Collaborating Action by Action: A Multi-agent LLM Framework for Embodied Reasoning

T0 review · 3 major / 8 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read This paper claims that for current LLM agents, the main barrier to embodied teamwork is inefficient natural-language communication, not missing task skill.

desk verdict MineCollab is a genuinely useful benchmark and the paper is clearly written, but the headline 'communication bottleneck' claim rests on ablations that change information access rather than communication cost. read the letter →

arxiv 2504.17950 v1 pith:QLGTLKQC submitted 2025-04-24 cs.MA cs.CL

Isadora White , Kolby Nottingham , Ayush Maniar , Max Robinson , Hansen Lillemark , Mehul Maheshwari , Lianhui Qin , Prithviraj Ammanabrolu This is my paper · ORCID
classification cs.MAcs.CL
keywords multi-agentcollaborationembodiedreasoninglargelanguagemodelsMinecraftnaturalcommunicationbenchmarksupervisedfine-tuning
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

Current large language models can perform individual embodied tasks, but this paper argues that when they must work together, the scarce resource is not skill—it is efficient natural-language communication. To study this, the authors build MINDcraft, a Minecraft platform that gives agents 47 high-level tools and pairwise chat, and MineCollab, a benchmark spanning cooking, crafting, and construction with teams of two to five agents. Across several leading models, success drops by more than 15% when teams are forced to transfer a detailed step-by-step plan, and scaling from two to five agents can lower success from about 90% to below 30%. The paper also shows that fine-tuning a small model on successful collaborative trajectories brings it close to a much larger model, which it reads as evidence that current prompting and imitation methods are not enough for collaboration.

What carries the argument

The load-bearing mechanism is MineCollab's plan-visibility ablation, implemented inside MINDcraft's agent architecture. MINDcraft gives each agent 47 parameterized tools, such as !givePlayer, !getCraftingPlan, and !checkBlueprintLevel, plus a pairwise conversation manager, so task failures cannot be blamed on low-level movement or syntax. The decisive conditions are Full Plan, where every agent sees the recipe; Partial Plan, where one agent must obtain it from a partner; and Hell's Kitchen, where each agent holds only part of the recipe; these are combined with team-size sweeps and blueprint-complexity sweeps to separate having information from being forced to communicate it.

What would settle it

Give every agent the full recipe in its prompt but restrict messages to a fixed small budget (or force strictly alternating one-at-a-time turns); if success stays equal to free-chat performance, communication cost is not the cause of the 15% drop. Alternatively, give one agent the recipe, allow unrestricted chat, then delete the recipe from that agent's memory mid-episode; if success collapses, the bottleneck is retained information, not transmission.

Watch

Extended reading notes

Core claim

The paper claims that existing LLM agents are ill-optimized for multi-agent collaboration, especially in embodied settings, and that the main bottleneck is efficient natural-language communication rather than lack of underlying task competence. The evidence is a set of controlled ablations: hiding the recipe or crafting plan from one agent and forcing the team to ask for and explain it lowers success across all tested models by over 15%; increasing team size from two to five agents drops success rates from up to 90% to under 30%; and in blueprint construction the best model still places less than 40% of required blocks. A related result is that an 8B model trained by behavior cloning on filtered successful runs of a 70B model roughly matches or exceeds the larger models on crafting and construction, suggesting the platform's data is usable for improving smaller agents.

Load-bearing premise

The paper's interpretation of the plan-visibility results depends on treating the hidden-recipe conditions as a measure of communication burden, but those conditions also withhold the recipe from one agent, so the performance drop could be caused by missing information itself rather than by the cost of transmitting it.

Editorial extensions

If this is right

  • If communication is the bottleneck, then improving the message layer—what agents say, when, and how briefly—should help current systems more than adding task-specific planning skills.
  • Naively adding agents hurts: the two-to-five-agent curves drop from up to 90% to under 30% success, so parallelization without coordination-load management is counterproductive.
  • Filtered successful trajectories are a usable training signal: a small model fine-tuned on them matches or beats much larger models on crafting and construction.
  • Plan visibility should be reported as a standard axis in collaboration benchmarks, because in several conditions it changes success more than the choice of model.

Reading between the lines

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

  • A sharper test of the communication-bottleneck claim would keep task information identical across agents and throttle only the channel—for example, capping message length or forcing strictly alternating turns; if success does not drop, the 15% effect is about missing knowledge, not communication cost.
  • The pairwise chat design implies coordination cost likely grows superlinearly with team size, so a broadcast or shared-blackboard protocol is a natural next experiment the paper does not run.
  • The hidden-plan design transfers to human-AI collaboration: one would predict that splitting instructions between a person and an agent and forcing verbal transfer degrades joint performance more than giving both parties the full instructions.
  • Because construction trajectories last over 100 steps on average and failures compound—agents undo each other's work—the communication penalty may be larger than 15% on long-horizon tasks even though the paper reports the main drop on shorter cooking and crafting tasks.
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

3 major / 8 minor

Summary. The paper introduces MINDcraft, a plugin-based Minecraft platform for LLM agents, and MineCollab, a benchmark of cooking, crafting, and construction tasks. It evaluates several closed and open LLMs (GPT-4o, Claude 3.5 Sonnet, LLaMA-3.3-70B-Instruct, LLaMA-3-8B, and an SFT variant) on these tasks, ablates agent count, plan visibility, and blueprint complexity, and trains an 8B model on oracle-generated SFT data. The authors' central claim is that natural-language communication is the primary bottleneck in multi-agent embodied collaboration, with performance dropping by up to 15% when agents must communicate detailed plans.

Significance. The platform and benchmark are timely and potentially widely useful: the procedural train/test split, the open-weight SFT pipeline, and the qualitative failure analyses are concrete contributions. If the communication-bottleneck claim were cleanly established, it would redirect research toward inter-agent communication efficiency rather than individual task competence. However, the claim currently rests on an information-access confound and on very small samples, so the paper's headline conclusion is not yet established.

major comments (3)
  1. [Section 5 (Figure 3c/3d; Tables 4 and 5)] The plan-visibility manipulation changes information access, not communication efficiency. In the crafting Partial Plan condition one agent is denied the crafting plan, and in the Hell's Kitchen variant agents lack recipes for required dishes; an agent without the plan cannot proceed until the missing content is transmitted. The observed performance drops therefore conflate the absence of information with the cost of communicating it. The pattern is also not monotonic in the way a communication-burden account predicts: Table 5 shows GPT-4o succeeding more when both agents are blocked (83.3%) than when one is blocked (58.3%), and Table 4 shows LLaMA-3.3-70B and LLaMA-8B-SFT improving slightly under Partial Plan (35.71%→36.36% and 42.86%→45.45%, respectively). Please add a condition that holds the information content constant and varies only the communication channel's capacity (e.g., message-length limits, added latency, or shared vs. private state) to isolate communication efficiency, or substantially soften the headline claim.
  2. [Tables 4-5 and Figure 3] The reported differences are within sampling noise. Per-cell sample sizes are n=14 and n=11 in Table 4, so the headline drops (Claude: 85.71%→72.73%; GPT-4o: 35.71%→27.27%) correspond to one or two trial flips. No confidence intervals, bootstrap estimates, or significance tests are reported for any ablation in Section 5, including Table 7. Please report per-cell trial counts and uncertainty measures, and interpret effect sizes accordingly.
  3. [Section 4, Table 3, Table 6, Figure 3e/3f] The construction evaluation metric is ambiguous. Section 4 defines construction scoring as an edit-distance metric and Table 3's caption calls the construction row 'the average of those edit distance scores,' yet Figure 3e/3f and Table 6 label the same construction evaluations as 'Success Rate,' and the Introduction interprets the construction numbers as 'percentage of blocks placed.' These are different quantities with opposite direction of goodness. Please state the precise metric, use consistent labels across all tables and figures, and report both the raw distance and a threshold-based success rate if both are intended.
minor comments (8)
  1. [Section 5] The phrase 'gold truth crafting plan' should be 'ground truth crafting plan.'
  2. [Table 5] Please report per-cell sample sizes for the No Agents Blocked / 1 Agent Blocked / Both Agents Blocked conditions.
  3. [Table 7] The caption says 'Using llama 70B' but the table does not state the number of trials per ablation; please add n and clarify whether this is the base LLaMA-3.3-70B-Instruct model.
  4. [Table 2] The 'Success' column is ambiguous; clarify that these counts are the number of successful runs retained for SFT construction, not success rates.
  5. [Section 3.1] The text states the tool library contains 47 parameterized tools, but Table 8 in Appendix 12 appears to list fewer; reconcile the count or note that some tools are omitted.
  6. [References] Reference [13] is also titled 'MindCraft'; the name collision with this paper's MINDcraft should be explicitly acknowledged.
  7. [Table 1] The 'Partial Obs' column is checked for all platforms, but the observation mechanisms differ widely (tool queries vs. full-state access); add a short definition in the caption or table notes.
  8. [Figure 3 caption] The caption says 'models drop by over 15% when forced to communicate these plans'; this should be rephrased to 'when denied access to these plans' to match the actual manipulation.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the results are empirical benchmark measurements and held-out SFT evaluations, with no load-bearing self-citation or definitional reduction.

full rationale

This paper contains no mathematical derivation whose conclusion is identical to its premises. The central results are empirical: performance numbers from LLM agents on MineCollab tasks, and the SFT model is trained on successful oracle trajectories from the train split and evaluated on held-out test recipes and blueprints (Section 4, Table 2), which is standard distillation rather than circular reasoning. The headline claim that communication is the primary bottleneck is an empirical interpretation of ablations that vary recipe and plan visibility (Section 5, Figure 3c/3d, Tables 4-5); regardless of whether those ablations cleanly isolate communication cost from information access, the claim is not true by construction and does not rename a fitted parameter or import a self-cited uniqueness theorem. The paper's self-citations (e.g., the MINDcraft software repository and Selective Perception [25]) are tooling or background context and are not load-bearing justifications for the headline result. No circular step is exhibited in the paper's own equations or evaluation protocol.

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

The central claims of this benchmark paper rest on a handful of domain assumptions about the validity of Minecraft as a proxy for embodied reasoning and about the interpretation of the ablation results. No mathematical axioms are involved. The free parameters are hyperparameters chosen by hand. No new physical or conceptual entities are introduced.

free parameters (3)
  • Memory summary interval = 15 steps
    Memory summarization is performed every 15 steps; the interval is a design choice that likely affects agent performance (Table 7 shows removing memory hurts).
  • Construction SFT data selection threshold = top 25%
    SFT data for construction is filtered to trials scoring in the top 25% by edit distance; the threshold is arbitrary and affects the training distribution.
  • Task timeout = 10 minutes + 5 per room complexity
    Task timeout is a hyperparameter that may influence success rates, especially for long-horizon construction.
assumptions (5)
  • domain assumption Minecraft is a valid surrogate for real-world embodied reasoning
    The paper uses Minecraft as the testbed; the validity of transferring conclusions to physical embodied agents is assumed, not demonstrated.
  • domain assumption Textual observations are sufficient for embodied reasoning performance
    Section 3.1 states that vision inputs were not rigorously evaluated and textual observations often outperform visual inputs, citing prior work [26]; this is assumed for the benchmark.
  • domain assumption Successful llama3.3-70b trajectories are high-quality supervision
    Section 4 uses successful llama3.3-70b runs as SFT data; this presumes the oracle's behavior is a good target for imitation.
  • domain assumption Train/test recipe and blueprint separation prevents memorization
    Section 4 states train/test tasks involve different recipes and blueprints; this assumes LLMs do not generalize through memorized Minecraft crafting knowledge.
  • domain assumption The plan visibility manipulation measures communication burden
    The hidden plan ablation (Hell's Kitchen, partial plan) is interpreted as imposing communication load, but it also changes information access; this is a key interpretive assumption.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Collaborating Action by Action: A Multi-agent LLM Framework for Embodied Reasoning." pith.science (2026). https://pith.science/paper/QLGTLKQC

@misc{pith2026250417950,
  author       = {Pith},
  title        = {Pith review of: Collaborating Action by Action: A Multi-agent LLM Framework for Embodied Reasoning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QLGTLKQC}},
  note         = {Machine review of arXiv:2504.17950}
}
read the original abstract

Collaboration is ubiquitous and essential in day-to-day life -- from exchanging ideas, to delegating tasks, to generating plans together. This work studies how LLMs can adaptively collaborate to perform complex embodied reasoning tasks. To this end we introduce MINDcraft, an easily extensible platform built to enable LLM agents to control characters in the open-world game of Minecraft; and MineCollab, a benchmark to test the different dimensions of embodied and collaborative reasoning. An experimental study finds that the primary bottleneck in collaborating effectively for current state-of-the-art agents is efficient natural language communication, with agent performance dropping as much as 15% when they are required to communicate detailed task completion plans. We conclude that existing LLM agents are ill-optimized for multi-agent collaboration, especially in embodied scenarios, and highlight the need to employ methods beyond in-context and imitation learning. Our website can be found here: https://mindcraft-minecollab.github.io/

Figures

Figures reproduced from arXiv: 2504.17950 by the authors.

Figure 1
Figure 1. Task suites and challenges. In this figure, we see the collaborative and embodied reasoning challenges displayed. In the cooking and crafting tasks, the agents need to delegate tasks, share resources and use embodied planning to manipulate the world of Minecraft. In the construction tasks, the agents need to navigate and coordinate in the space to ensure they consistently build towards their objective without undoin… view at source ↗
Figure 2
Figure 2. Overview of the MINDcraft workflow. A user or task configuration (left) provides instruc￾tions (e.g., “Build a house out of nearby materials”). The Agent (center) takes these instructions, consults an LLM (via a model request) and invokes high-level commands/tools. These commands are then executed in the Minecraft environment (right), with the agent receiving feedback through execution logs. The extensive command li… view at source ↗
Figure 3
Figure 3. Task complexity ablations. In the first row, we ablate different numbers of agents in the crafting and cooking tasks. Construction tasks can also be run with 3+ agent tasks, but are outside of our budget for closed source APIs. In the second row, we ablate access to hidden plan information like the recipe for a cake (cooking) or the steps to make a bookshelf (crafting) find that models drop by over 15% when forced t… view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 4 Pith papers

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

  1. Diagon: A Programmable Testbed for AI-Agent Cognitive Labor Markets

    cs.CE 2026-04 unverdicted novelty 7.0 of 10

    Market exchange among AI agents can raise productivity over self-sufficient agents, but institutional rules such as identity transparency and stronger selection can degrade those gains.

  2. GuessBench: Sensemaking Multimodal Creativity in the Wild

    cs.CL 2025-06 conditional novelty 7.0 of 10

    A Minecraft-based benchmark shows vision-language models often fail to decode player-built creations, with accuracy falling sharply for rare concepts and low-resource languages.

  3. Structured LLM Reasoning for Zero-Shot Human--Robot Coordination Under Hidden Goals

    cs.RO 2026-08 conditional novelty 5.0 of 10

    A Dec-POMDP-structured LLM architecture with theory-of-mind inference improved coordination steps and trust in a five-person pilot, though the evidence is limited.

  4. Personalized Constitutionally-Aligned Agentic Superego: Secure AI Behavior Aligned to Diverse Human Values

    cs.AI 2025-06 conditional novelty 5.0 of 10

    An external 'superego' module that filters agentic AI plans against user-selected 'constitutions' plus a universal safety floor is reported to cut harmful outputs by up to 98% on safety benchmarks.

Reference graph

Works this paper leans on

34 extracted references · 19 canonical work pages · cited by 4 Pith papers

  1. [1]

    Why do multi- agent llm systems fail? arXiv preprint arXiv:2503.13657, 2025

    Mert Cemri, Melissa Z Pan, Shuyi Yang, Lakshya A Agrawal, Bhavya Chopra, Rishabh Tiwari, Kurt Keutzer, Aditya Parameswaran, Dan Klein, Kannan Ramchandran, et al. Why do multi- agent llm systems fail? arXiv preprint arXiv:2503.13657, 2025

    arXiv 2025

  2. [2]

    Large language model based multi-agents: A survey of progress and challenges

    Taicheng Guo, Xiuying Chen, Yaqi Wang, Ruidi Chang, Shichao Pei, Nitesh V Chawla, Olaf Wiest, and Xiangliang Zhang. Large language model based multi-agents: A survey of progress and challenges. arXiv preprint arXiv:2402.01680, 2024

    arXiv 2024

  3. [3]

    On the utility of learning about humans for human-ai coordination

    Micah Carroll, Rohin Shah, Mark K Ho, Tom Griffiths, Sanjit Seshia, Pieter Abbeel, and Anca Dragan. On the utility of learning about humans for human-ai coordination. Advances in neural information processing systems, 32, 2019

    2019

  4. [4]

    Executing instructions in situated collaborative interactions

    Alane Suhr, Claudia Yan, Jack Schluger, Stanley Yu, Hadi Khader, Marwa Mouallem, Iris Zhang, and Yoav Artzi. Executing instructions in situated collaborative interactions. In Kentaro Inui, Jing Jiang, Vincent Ng, and Xiaojun Wan, editors, Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint...

    work page 2019

  5. [5]

    Habitat: A platform for embodied ai research

    Manolis Savva, Abhishek Kadian, Oleksandr Maksymets, Yili Zhao, Erik Wijmans, Bhavana Jain, Julian Straub, Jia Liu, Vladlen Koltun, Jitendra Malik, et al. Habitat: A platform for embodied ai research. In Proceedings of the IEEE/CVF international conference on computer vision, pages 9339–9347, 2019

    2019

  6. [6]

    Llm-coordination: evaluating and analyzing multi-agent coordination abilities in large language models

    Saaket Agashe, Yue Fan, Anthony Reyna, and Xin Eric Wang. Llm-coordination: evaluating and analyzing multi-agent coordination abilities in large language models. arXiv preprint arXiv:2310.03903, 2023

    arXiv 2023

  7. [7]

    Generative agents: Interactive simulacra of human behavior

    Joon Sung Park, Joseph O’Brien, Carrie Jun Cai, Meredith Ringel Morris, Percy Liang, and Michael S Bernstein. Generative agents: Interactive simulacra of human behavior. In Proceed- ings of the 36th annual acm symposium on user interface software and technology, pages 1–22, 2023

    2023

  8. [8]

    Partnr: A benchmark for planning and reasoning in embodied multi-agent tasks

    Matthew Chang, Gunjan Chhablani, Alexander Clegg, Mikael Dallaire Cote, Ruta Desai, Michal Hlavac, Vladimir Karashchuk, Jacob Krantz, Roozbeh Mottaghi, Priyam Parashar, et al. Partnr: A benchmark for planning and reasoning in embodied multi-agent tasks. arXiv preprint arXiv:2411.00081, 2024

    arXiv 2024

Show all 34 references
  1. [9]

    Mineland: Simulating large-scale multi- agent interactions with limited multimodal senses and physical needs

    Xianhao Yu, Jiaqi Fu, Renjia Deng, and Wenjuan Han. Mineland: Simulating large-scale multi- agent interactions with limited multimodal senses and physical needs. CoRR, abs/2403.19267, 2024

    2024 arXiv

  2. [10]

    Mastering diverse control tasks through world models

    Danijar Hafner, Jurgis Pasukonis, Jimmy Ba, and Timothy Lillicrap. Mastering diverse control tasks through world models. Nature, pages 1–7, 2025

    2025

  3. [11]

    V oyager: An open-ended embodied agent with large language models

    Guanzhi Wang, Yuqi Xie, Yunfan Jiang, Ajay Mandlekar, Chaowei Xiao, Yuke Zhu, Linxi Fan, and Anima Anandkumar. V oyager: An open-ended embodied agent with large language models. Transactions on Machine Learning Research, 2023

    2023

  4. [12]

    Do embodied agents dream of pixelated sheep: Embodied decision making using language guided world modelling

    Kolby Nottingham, Prithviraj Ammanabrolu, Alane Suhr, Yejin Choi, Hannaneh Hajishirzi, Sameer Singh, and Roy Fox. Do embodied agents dream of pixelated sheep: Embodied decision making using language guided world modelling. In International Conference on Machine Learning, pages...

    2023

  5. [13]

    MindCraft: Theory of mind modeling for situated dialogue in collaborative tasks

    Cristian-Paul Bara, Sky CH-Wang, and Joyce Chai. MindCraft: Theory of mind modeling for situated dialogue in collaborative tasks. In Marie-Francine Moens, Xuanjing Huang, Lucia Specia, and Scott Wen-tau Yih, editors, Proceedings of the 2021 Conference on Empirical Methods in N...

    2021

  6. [14]

    Cooperate or collapse: Emergence of sustainable cooperation in a society of llm agents

    Giorgio Piatti, Zhijing Jin, Max Kleiman-Weiner, Bernhard Schölkopf, Mrinmaya Sachan, and Rada Mihalcea. Cooperate or collapse: Emergence of sustainable cooperation in a society of llm agents. In The Thirty-eighth Annual Conference on Neural Information Processing Systems, 2024

    2024

  7. [15]

    is this it?

    Dan Bohus, Sean Andrist, Yuwei Bao, Eric Horvitz, and Ann Paradiso. " is this it?": Towards ecologically valid benchmarks for situated collaboration. In Companion Proceedings of the 26th International Conference on Multimodal Interaction, pages 41–45, 2024

    2024

  8. [16]

    Infer human’s intentions before following natural language instructions

    Yanming Wan, Yue Wu, Yiping Wang, Jiayuan Mao, and Natasha Jaques. Infer human’s intentions before following natural language instructions. arXiv preprint arXiv:2409.18073, 2024

    2024 arXiv

  9. [17]

    Cps- taskforge: Generating collaborative problem solving environments for diverse communication tasks

    Nikita Haduong, Irene Wang, Bo-Ru Lu, Prithviraj Ammanabrolu, and Noah A Smith. Cps- taskforge: Generating collaborative problem solving environments for diverse communication tasks. arXiv preprint arXiv:2408.08853, 2024

    2024 arXiv

  10. [18]

    Embodied agent interface: Benchmarking llms for embodied decision making

    Manling Li, Shiyu Zhao, Qineng Wang, Kangrui Wang, Yu Zhou, Sanjana Srivastava, Cem Gokmen, Tony Lee, Erran Li Li, Ruohan Zhang, et al. Embodied agent interface: Benchmarking llms for embodied decision making. Advances in Neural Information Processing Systems , 37:100428–100534, 2024

    2024

  11. [19]

    Procthor: Large-scale embodied ai using procedural generation

    Matt Deitke, Eli VanderBilt, Alvaro Herrasti, Luca Weihs, Kiana Ehsani, Jordi Salvador, Winson Han, Eric Kolve, Aniruddha Kembhavi, and Roozbeh Mottaghi. Procthor: Large-scale embodied ai using procedural generation. Advances in Neural Information Processing Systems, 35:5982–5...

    2022

  12. [20]

    Building cooperative embodied agents modularly with large language models

    Hongxin Zhang, Weihua Du, Jiaming Shan, Qinhong Zhou, Yilun Du, Joshua B Tenenbaum, Tianmin Shu, and Chuang Gan. Building cooperative embodied agents modularly with large language models. In The Twelfth International Conference on Learning Representations, 2023

    2023

  13. [21]

    Teach: Task-driven embodied agents that chat

    Aishwarya Padmakumar, Jesse Thomason, Ayush Shrivastava, Patrick Lange, Anjali Narayan- Chen, Spandana Gella, Robinson Piramuthu, Gokhan Tur, and Dilek Hakkani-Tur. Teach: Task-driven embodied agents that chat. In Proceedings of the AAAI Conference on Artificial Intelligence, ...

    2017

  14. [22]

    Optima: Optimizing effectiveness and efficiency for llm-based multi-agent system

    Weize Chen, Jiarui Yuan, Chen Qian, Cheng Yang, Zhiyuan Liu, and Maosong Sun. Optima: Optimizing effectiveness and efficiency for llm-based multi-agent system. arXiv preprint arXiv:2410.08115, 2024

    2024 arXiv

  15. [23]

    Malt: Improving reasoning with multi-agent llm training

    Sumeet Ramesh Motwani, Chandler Smith, Rocktim Jyoti Das, Rafael Rafailov, Ivan Laptev, Philip HS Torr, Fabio Pizzati, Ronald Clark, and Christian Schroeder de Witt. Malt: Improving reasoning with multi-agent llm training. arXiv preprint arXiv:2412.01928, 2024

    2024

  16. [24]

    Magdi: structured distillation of multi-agent interaction graphs improves reasoning in smaller language models

    Justin Chih-Yao Chen, Swarnadeep Saha, Elias Stengel-Eskin, and Mohit Bansal. Magdi: structured distillation of multi-agent interaction graphs improves reasoning in smaller language models. In Proceedings of the 41st International Conference on Machine Learning , pages 7220–7235, 2024

    2024

  17. [25]

    Selective perception: Learning concise state descriptions for language model actors

    Kolby Nottingham, Yasaman Razeghi, Kyungmin Kim, Jb Lanier, Pierre Baldi, Roy Fox, and Sameer Singh. Selective perception: Learning concise state descriptions for language model actors. In Kevin Duh, Helena Gomez, and Steven Bethard, editors, Proceedings of the 2024 Conference...

    2024

  18. [26]

    Mind’s eye of llms: visualization-of-thought elicits spatial reasoning in large language models

    Wenshan Wu, Shaoguang Mao, Yadong Zhang, Yan Xia, Li Dong, Lei Cui, and Furu Wei. Mind’s eye of llms: visualization-of-thought elicits spatial reasoning in large language models. Advances in Neural Information Processing Systems, 37:90277–90317, 2025

    2025

  19. [27]

    Minerl diamond 2021 competition: Overview, results, and lessons learned

    Anssi Kanervisto, Stephanie Milani, Karolis Ramanauskas, Nicholay Topin, Zichuan Lin, Jun- you Li, Jianing Shi, Deheng Ye, Qiang Fu, Wei Yang, et al. Minerl diamond 2021 competition: Overview, results, and lessons learned. NeurIPS 2021 Competitions and Demonstrations Track, pa...

    2021

  20. [28]

    Video pretraining (vpt): Learning to act by watching unlabeled online videos

    Bowen Baker, Ilge Akkaya, Peter Zhokov, Joost Huizinga, Jie Tang, Adrien Ecoffet, Brandon Houghton, Raul Sampedro, and Jeff Clune. Video pretraining (vpt): Learning to act by watching unlabeled online videos. Advances in Neural Information Processing Systems, 35:24639–24654, 2022

    2022

  21. [29]

    Mineflayer

    PrismarineJS. Mineflayer. MIT License

  22. [30]

    Retrieval-augmented generation for knowledge-intensive nlp tasks

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

    2020

  23. [31]

    Why greatness cannot be planned: The myth of the objective

    Kenneth O Stanley and Joel Lehman. Why greatness cannot be planned: The myth of the objective. (No Title), 2015

    2015

  24. [32]

    terrance

    Prithviraj Ammanabrolu, Jack Urbanek, Margaret Li, Arthur Szlam, Tim Rocktäschel, and Jason Weston. How to motivate your dragon: Teaching goal-driven agents to speak and act in fantasy worlds. In Kristina Toutanova, Anna Rumshisky, Luke Zettlemoyer, Dilek Hakkani-Tur, Iz Belta...

    2021

  25. [33]

    Toggling the building assistance feature by setting the cheat variable to true in the task_construction profile

  26. [34]

    Response:

    Accessing the generateConstructionTasks function in generate_multiagent_construction_tasks.js to implement custom complex- ity levels beyond the predefined parameters. The following can be changed here: (a) Room size, window / carpet style (b) Number of variants (c) Timeout du...

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.