Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

Persistent workspaces and reusable skills are the mechanism that turns chatbots into durable digital colleagues.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-02 11:25 UTC pith:WCZZYTPL

load-bearing objection A useful synthesis of the agent-to-workspace trend, but the central causal claim — Workspace+Skill is the key leap — is asserted, not demonstrated. the 3 major comments →

arxiv 2606.14502 v2 pith:WCZZYTPL submitted 2026-06-12 cs.AI

From Chatbot to Digital Colleague: The Paradigm Shift Toward Persistent Autonomous AI

Yongheng Zhang , Ziang Liu , Jiaxuan Zhu , Shuai Wang , Xiangqi Chen , Haojing Huang , Jiayi Kuang , Siyu Chen
show 12 more authors
Ao Shen Hao Wu Qiufeng Wang Qian-Wen Zhang Junnan Dong Wenhao Jiang Ying Shen Hai-Tao Zheng Yinghui Li Di Yin Xing Sun Philip S. Yu
This is my paper
classification cs.AI
keywords LLM agentsdigital colleagueworkspaceskillstask closureautonomous AIthinking LLMsagent evaluation
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This survey argues that the shift from chatbot to digital colleague is not primarily about smarter single-pass language models. The key leap, it claims, is combining a persistent Workspace—files, terminals, browsers, logs, permissions—with reusable Skills, packaged procedures that encode how to do a class of task. Together these provide state, memory, evidence, error recovery, and verification, turning isolated tool calls into closed, auditable work. The paper organizes the field along two dimensions: the cognitive core from fast response to deliberate reasoning, and tool execution from ad-hoc agents to workstation systems. If the thesis is right, progress on reliable autonomous AI depends as much on harness engineering, task-closure evaluation, and governance as on model scaling.

Core claim

The paper's central claim is that Workspace + Skill is the decisive architectural step. A Workspace is a persistent digital environment where files, terminals, browsers, repositories, logs, and permissions survive across a task; a Skill is a reusable, parameterizable procedure with instructions, scripts, checks, dependencies, and safety constraints. Together they convert episodic, best-effort tool use into persistent, inspectable work: the agent can load a procedure, operate on durable state, detect and repair failures, and leave a verified final workspace state. The authors assert that current agent failures—fragmented perception, ephemeral tool calls, brittleness under environmental noise,

What carries the argument

The central mechanism is the pair Workspace + Skill. The Workspace supplies persistent state and evidence—files, terminals, browsers, logs, permissions, snapshots—so that actions have inspectable and recoverable consequences. The Skill supplies procedural memory—packaged instructions, scripts, validation checks, dependencies, and safety constraints—so that repeated work does not have to be rediscovered each time. The paper argues that only when both are present does an agent achieve task closure: reaching and verifying the intended final state under reproducible and safe conditions. Workstation-style agent systems are presented as the representative engineering form of this mechanism.

Load-bearing premise

The load-bearing premise is that today's agent failures—fragmented perception, ephemeral tool calls, brittleness, missing task closure—are a fundamental architectural limitation of the environment-action-feedback loop, rather than simply a shortfall in model capability, training, or reasoning; if long-horizon unreliability is mostly a model-cognition problem, then persistent workspaces and skills are helpful scaffolding but not the decisive leap.

What would settle it

Run a controlled comparison of the same base model and agent loop on a long-horizon stateful benchmark (for example, desktop or web tasks with execution-based checks) under three conditions: stateless tool calls, a persistent workspace without skills, and persistent workspace plus a skill library. If task-closure rates do not improve materially when workspace state and skills are added—while the model is held fixed—the survey's central claim is falsified. A weaker disconfirmation would be showing that model scaling alone reproduces the same reliability gains without any workspace changes.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • If Workspace + Skill is the key leap, then the binding constraint on reliable autonomous AI is the execution substrate—state persistence, verification loops, permissions, rollback—alongside the model's reasoning ability.
  • Agent training data should be built from complete state-action-observation trajectories, including tool outputs, intermediate failures, and final-state evidence, rather than static instruction-response pairs.
  • Evaluation should move to task closure: final-state verification, repeated-run reliability, efficiency, reproducibility, and trajectory-level safety, instead of answer-level accuracy.
  • The main bottleneck in deploying agents shifts from prompt design to system operations: skill lifecycle management, workspace hygiene, sandboxing, audit trails, and governance.
  • Human-AI interaction shifts from instruction-following to delegation—users set objectives, constraints, permissions, and acceptance criteria, then audit the work episode.

Where Pith is reading between the lines

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

  • Editorial inference: If the architectural thesis is right, harness quality may matter more than model scale for practical long-horizon work; a well-instrumented workspace could let smaller, cheaper models compete with much larger ones on real tasks.
  • Testable extension: A controlled ablation—same base model and instruction set, run with and without persistent workspace state and a reusable skill library on a stateful benchmark—would isolate whether the gains attributed to Workspace + Skill are architectural or just extra context and tool access.
  • Neighbouring consequence: The skill-as-package view predicts that skill provenance, versioning, and dependency checking become as important as model safety, and that supply-chain attacks on skill libraries will be a primary failure mode.
  • The delegation framing implies that research on AI interfaces should focus on authority, escalation, and audit surfaces rather than chat alone; progress may be measured by how little human micro-management is needed at a given level of risk.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. This paper is a broad survey and position paper that organizes recent LLM evolution along two dimensions: the cognitive core (Chatbot → Thinking LLM) and tool-augmented task execution (Agent → OpenClaw-style workstation systems). Its central claim, stated in the Introduction and developed in Part III, is that the combination of a persistent Workspace and reusable Skills is "the key leap" that transforms chatbot-style interaction into durable, task-closing "digital colleague" work. The paper reviews models, benchmarks, data regimes, and evaluation methods across four eras, and it discusses open challenges in reliability, governance, memory, and self-evolving ecosystems. It is explicitly a synthesis rather than a new experimental study: no experiments are run, and the framework is assembled from existing systems and benchmarks.

Significance. If the central thesis is accepted, the paper makes a useful conceptual contribution by shifting attention from model-scale and reasoning ability alone to the execution substrate, skill libraries, verification loops, and governance mechanisms that enable long-horizon task closure. The survey's strengths include its broad and current coverage of systems and benchmarks, the clear two-dimensional framing, the concrete taxonomy of data and evaluation stages (Tables 6–11), and a candid list of limitations of the Workspace + Skill paradigm in §4.2.2. It also usefully connects technical reliability with security, forensics, and organizational governance. The paper does not ship machine-checked proofs or code, but it does provide a falsifiable framing: the claim that persistent workspaces and reusable skills are causally important could, in principle, be tested by controlled comparisons. The main weakness is that the load-bearing causal attribution is asserted rather than evidenced.

major comments (3)
  1. [§3.1.2 and Part III] The paper's central claim that "Workspace + Skill is the key leap" is a causal attribution that is not tested. §3.1.2 calls the four Agent-era bottlenecks "a fundamental architectural limitation" rather than a reflection of insufficient model capability, and Part III builds on this. However, no controlled comparison separates the effect of the workspace/skill substrate from model cognition: the cited benchmarks (WebArena, SWE-bench, OSWorld) compare different models or settings, and Table 4's Agent/OpenClaw boundary is a definitional dichotomy, not empirical evidence. The failure modes listed are equally consistent with limited planning, reasoning, and self-correction in the base model. To make the central claim defensible, the paper should either reframe it as a proposal/hypothesis with explicit testable predictions, or present the available evidence in a way that separates substrate ef
  2. [Figure 2] The figure claims that "the time horizon of frontier AI agents has grown exponentially" and presents this as a key takeaway. Yet no fitted curve, confidence interval, or regression is shown, and the provenance of the underlying "50%-time horizon" data is only a footnote to an external website. Axis units are mixed (seconds in one label, minutes in another), and the methodology for computing the median task length is not described. If this exponential claim is load-bearing for the paper's narrative, the data points and fitting procedure should be reported; otherwise the claim should be softened to "approximately exponential in the observed period" or removed.
  3. [§4.2.2 and Conclusion] The paper's own limitation list — skill brittleness, environmental drift, negative transfer, workspace contamination, security/supply-chain risk, and governance overhead — substantially weakens the "key leap" framing. These are not merely operational details; they show that the benefits of Workspace + Skill are conditional on an expensive governance and maintenance layer. The manuscript should state under which conditions the paradigm is decisive (e.g., bounded, versioned environments with strong verification) and where it acts only as scaffolding atop model capability. Without this, Part III's conclusion overreaches relative to the evidence the paper itself presents.
minor comments (5)
  1. [Section 1] Typo: "the central question is thereforeno longer limited tohow can a model generate a better answer?Instead, it is howhow can an AI system reliably transform user intent into completed work?" — "howhow" should read "how".
  2. [Figure 1] Several node labels contain typos or inconsistent formatting: "Qwen3-Instuct" should be "Qwen3-Instruct", "Dep2025" is likely "Dec2025", and the legend text about open/closed box styles is missing a glyph. Please also ensure the timeline dates are consistent between text and figure.
  3. [Figure 2] The caption refers to a footnote for the data source, but the definition of "50%-time horizon" should be in the caption itself, along with the unit of measurement (seconds/minutes). The y-axis labels mix seconds and minutes, which makes the plot hard to read.
  4. [Table 10 notes] The note "UI-TARS-2 scores marked with 'use the paper's extended GUI-SDK setting" has an unmatched quotation mark. Also, "Terminal 2.0" is used as a column heading but the text refers to "Terminal-Bench v2.0"; this shorthand should be defined in the table notes.
  5. [§5.2.4 / Table 11] The selection rationale "retained columns were selected using Semantic Scholar citation-overlap" is not a transparent criterion. Either describe the exact selection procedure or report the full set of benchmark columns; otherwise the table may appear cherry-picked.

Circularity Check

0 steps flagged

No significant circularity: the Workspace+Skill thesis is a conceptual organizing claim, not a derived result, and the only self-citations are informational table entries.

full rationale

Walked the claimed derivation chain: the paper's central claim is that 'Workspace + Skill provides the mechanism that turns chatbot-style interaction into durable digital-colleague work' (Section 1). This is a survey thesis and organizing framework, not a formal derivation with fitted parameters or equations that reduce to its own inputs. The four Agent-era bottlenecks (§3.1.2) are attributed to external benchmarks and failure analyses (AgentBench, WebArena, SWE-bench, GAIA, LLM Agent Failure); the sentence calling them 'a fundamental architectural limitation' is a causal assertion, not a result derived within the paper, so it is a correctness-risk concern rather than a circularity. The paper's own §4.2.2 limitations paragraph explicitly concedes skill brittleness, environmental drift, negative transfer, workspace contamination, and supply-chain risk, which weakens the strength of the central claim but does not make it circular. The only author-affiliated items are Youtu-LLM and Youtu-VL in Table 1, which are purely informational timeline entries and are not cited anywhere in the load-bearing argument. The framework is assembled from external systems and benchmarks (OpenHands, SWE-agent, SWE-bench, WebArena, OSWorld, ClawsBench, etc.), and OpenClaw is used as a representative case study rather than as an input to a derivation. No circular step was found.

Axiom & Free-Parameter Ledger

0 free parameters · 3 axioms · 2 invented entities

The paper does not fit parameters, but its narrative rests on three unproved premises: the four-era taxonomy, the causal priority of Workspace+Skill over model capability, and the trustworthiness of copied benchmark tables. The conceptual labels 'Digital Colleague' and 'Workspace+Skill paradigm' are introduced without falsifiable predictions.

axioms (3)
  • domain assumption LLM evolution is separable into two dimensions (cognitive core; tool-augmented execution) and four eras (Chatbot, Thinking LLM, Agent, OpenClaw).
    The entire survey structure and Figure 1 depend on this taxonomy, but no evidence shows these axes are natural or exhaustive (§1–§3).
  • ad hoc to paper Persistent state (Workspace) and reusable procedures (Skill) are the decisive mechanism—the 'key leap'—for reliable long-horizon task closure.
    This is the paper's central thesis stated in §1 and Part III; it is asserted without a controlled comparison against model-scale or RL-based explanations.
  • domain assumption The benchmark scores and system facts in Tables 9–11 and Figure 2 are accurate and current as of submission.
    The empirical grounding consists of external leaderboards and vendor reports; the paper ships no data, retrieval scripts, or verification.
invented entities (2)
  • Digital Colleague (the 'next paradigm' state) no independent evidence
    purpose: Labels the target condition where AI systems perform persistent, governed work in shared workspaces.
    Introduced as a framing construct in §1; no measurable prediction or testable property distinguishes a Digital Colleague from a capable agent.
  • Workspace + Skill paradigm no independent evidence
    purpose: Proposed mechanism that supposedly converts episodic tool use into durable task closure.
    Defined in §4 as the key leap; no falsifiable hypothesis or experiment is provided, so it functions as an organizing label rather than an empirically grounded entity.

pith-pipeline@v1.3.0-alltime-deepseek · 50446 in / 13402 out tokens · 134347 ms · 2026-08-02T11:25:56.717673+00:00 · methodology

0 comments
read the original abstract

Large Language Models (LLMs) are undergoing a fundamental transformation from conversational generators into integrated AI systems capable of reasoning, action, memory, and self-improvement. We conceptualize this transition as a shift from Chatbot to Digital Colleague: from conversational answers to persistent work. We organize this transition along two tightly coupled dimensions. First, at the cognitive core level, LLMs are advancing from Chatbot-era "fast thinking" systems driven by next-token prediction toward Thinking LLMs that leverage inference-time computation, Chain-of-Thought reasoning, reflection, process supervision, and reinforcement learning to support more deliberate and reliable cognition. Second, at the tool-augmented task execution level, LLMs are progressing from tool-calling Agents that invoke external resources in an ad hoc manner toward OpenClaw-style workstation systems (OpenClaw) equipped with persistent Workspaces, skills, verification loops, and governance. The "Workspace + Skill" paradigm makes episodic tool use colleague-like via state persistence, reusable procedures, task closure, and experience reuse. We examine data construction shifts from instruction-response pairs to State-Action-Observation trajectories and evaluation from static benchmarks to sandboxed, auditable, self-evolving AI ecosystems.

Figures

Figures reproduced from arXiv: 2606.14502 by Ao Shen, Di Yin, Hai-Tao Zheng, Haojing Huang, Hao Wu, Jiaxuan Zhu, Jiayi Kuang, Junnan Dong, Philip S. Yu, Qian-Wen Zhang, Qiufeng Wang, Shuai Wang, Siyu Chen, Wenhao Jiang, Xiangqi Chen, Xing Sun, Yinghui Li, Ying Shen, Yongheng Zhang, Ziang Liu.

Figure 1
Figure 1. Figure 1: A roadmap and evolutionary timeline of next-generation LLM systems. The figure summarizes how these AI systems progress from simple conversational chatbots to reasoning cores, tool-using agents, and persistent workspace systems over time. Each node is labeled by its release month. box represents open-source / open platform; box represents closed / commercial system. 4 [PITH_FULL_IMAGE:figures/full_fig_p00… view at source ↗
Figure 2
Figure 2. Figure 2: Time horizon growth of frontier AI agents. Each point reports the 50%-time horizon, i.e., the median length of coding tasks that an agent can complete at release. The trend shows a transition from second-level fast-response models to slow-thinking models capable of sustaining increasingly long and complex tasks1 . 2.1 The Chatbot Era: Language Generation and Knowledge Compression Represented by ChatGPT Dur… view at source ↗
Figure 3
Figure 3. Figure 3: The Chatbot Era: a user inputs a natural-language question, the LLM performs fast, stateless, single-pass processing over compressed parametric knowledge, and immediately returns a fluent response. The figure highlights one-pass inference with no external loop, feedback-driven correction, or persistent memory. foundation for large-scale parallel training. Building on this architecture, the field establishe… view at source ↗
Figure 4
Figure 4. Figure 4: The Thinking LLM Era: the model allocates additional inference-time computation, generates long reasoning traces, explores alternatives, verifies intermediate steps, and then returns a more deliberate answer. The figure contrasts slow, reflective System-2-style reasoning with the chatbot’s fast single-pass response. shift. Before emitting a final response, these Reasoning LLMs (RLLMs) generate an extended … view at source ↗
Figure 5
Figure 5. Figure 5: The Agent Era: the model observes an external environment, plans the next step, invokes tools or actions, receives feedback, and iterates toward the task goal. The figure illustrates the observe–think–act– observe loop that gives LLMs an initial form of autonomy beyond single-turn answering. feedback, and iterating. This closed loop of observing, thinking, acting, and observing again is what fundamentally … view at source ↗
Figure 6
Figure 6. Figure 6: The OpenClaw Era: the agent works inside a persistent workspace with files, terminals, browsers, logs, permissions, reusable skills, and verification loops. The figure illustrates how workspace state and skill-based execution turn fragmented tool use into inspectable, recoverable, and deliverable task closure. 3.2 The OpenClaw Era: Persistent Workspaces for Task Closure Workspace intelligence: workspace ho… view at source ↗
Figure 7
Figure 7. Figure 7: Simple tool invocation: the LLM can call external tools to handle local sub-tasks, but these calls remain limited when the task requires persistent files, terminal sessions, execution logs, intermediate artifacts, and recoverable state. The figure highlights why a workspace is needed to support more complex, long-horizon task completion beyond isolated tool calls. 4.1.1 From Ephemeral Tool Calls to Persist… view at source ↗
Figure 8
Figure 8. Figure 8: Workspace + Skill paradigm: persistent workspaces provide the stateful place where work happens, while skills package reusable procedures, scripts, checks, and safety constraints. The figure shows how agents combine workspace context with skill assets to produce verifiable digital work instead of one-off responses. 4.2 Skills for Workspace Agents: Reusable Procedures for Repeatable Work Procedural memory f… view at source ↗
Figure 9
Figure 9. Figure 9: Data paradigm shift: training and evaluation data evolve from static prompt–response pairs to reasoning traces and state–action–observation trajectories. The figure shows why agentic and OpenClaw-style systems require tool outputs, UI states, workspace changes, and final-state evidence rather than only labels. 5.1 Data Paradigm Shift: From “Knowledge Corpus” to “Action Trajectory” From prompt–response pair… view at source ↗
Figure 10
Figure 10. Figure 10: Evaluation paradigm shift: evaluation moves from final-answer correctness to process judgment and task closure. The figure summarizes how next-generation systems must be assessed by reasoning validity, environment state changes, reliability, efficiency, reproducibility, and safety [PITH_FULL_IMAGE:figures/full_fig_p043_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Open challenges for reliable autonomy: as agents move from answering to acting in workspaces, failures become longer-horizon, stateful, and harder to reverse. The figure summarizes key bottlenecks around task closure, safety and governance, memory, context management, and persistent workspace state. state is achieved. Long-horizon tasks introduce several sources of instability. Errors can propagate across… view at source ↗
Figure 12
Figure 12. Figure 12: Future directions toward self-evolving AI ecosystems: next-generation systems will combine models, contexts, tools, skills, workspaces, memories, evaluators, and governance mechanisms into an integrated learning loop. The figure illustrates the path from reactive chatbots to governed digital colleagues that accumulate experience and improve their operating environments. 6.2.1 From Prompt Engineering to Ha… view at source ↗

discussion (0)

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

Forward citations

Cited by 1 Pith paper

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

  1. Thinking in Video: Can Video Generators Really Reason About the Real World?

    cs.CV 2026-07 conditional novelty 6.0

    Video generators show a perception-prediction gap: they can generate plausible continuations while failing explicit visual reasoning tests.

Reference graph

Works this paper leans on

296 extracted references · 115 linked inside Pith · cited by 1 Pith paper

  1. [1]

    On the opportunities and risks of foundation models.arXiv preprint arXiv:2108.07258, 2021

    Rishi Bommasani, Drew A Hudson, Ehsan Adeli, Russ Altman, Simran Arora, Sydney von Arx, Michael S Bernstein, Jeannette Bohg, Antoine Bosselut, Emma Brunskill, et al. On the opportunities and risks of foundation models.arXiv preprint arXiv:2108.07258, 2021

    Pith/arXiv arXiv 2021

  2. [2]

    Recent advances in natural language processing via large pre-trained language models: A survey.ACM Computing Surveys, 56(2): 1–40, 2023

    Bonan Min, Hayley Ross, Elior Sulem, Amir Pouran Ben Veyseh, Thien Huu Nguyen, Oscar Sainz, Eneko Agirre, Ilana Heintz, and Dan Roth. Recent advances in natural language processing via large pre-trained language models: A survey.ACM Computing Surveys, 56(2): 1–40, 2023

    2023

  3. [3]

    A comprehensive overview of large language models.ACM Transactions on Intelligent Systems and Technology, 16(5):1–72, 2025

    Humza Naveed, Asad Ullah Khan, Shi Qiu, Muhammad Saqib, Saeed Anwar, Muhammad Usman, Naveed Akhtar, Nick Barnes, and Ajmal Mian. A comprehensive overview of large language models.ACM Transactions on Intelligent Systems and Technology, 16(5):1–72, 2025

    2025

  4. [4]

    A survey on large language model based autonomous agents.Frontiers of Computer Science, 18(6), 2024

    Lei Wang, Chen Ma, Xueyang Feng, Zeyu Zhang, Hao Yang, Jingsen Zhang, Zhiyuan Chen, Jiakai Tang, Xu Chen, Yankai Lin, et al. A survey on large language model based autonomous agents.Frontiers of Computer Science, 18(6), 2024

    2024

  5. [5]

    The rise and potential of large language model based agents: A survey.arXiv preprint arXiv:2309.07864, 2023

    Zhiheng Xi, Wenxiang Chen, Xin Guo, Wei He, Yiwen Ding, Boyang Hong, Ming Zhang, Junzhe Wang, Senjie Jin, Enyu Zhou, et al. The rise and potential of large language model based agents: A survey.arXiv preprint arXiv:2309.07864, 2023

    Pith/arXiv arXiv 2023

  6. [6]

    Libo Qin, Qiguang Chen, Xiachong Feng, Yang Wu, Yongheng Zhang, Yinghui Li, Min Li, Wanxiang Che, and Philip S. Yu. Large language models meet nlp: A survey, 2025. URL https://arxiv.org/abs/2405.12819

    Pith/arXiv arXiv 2025

  7. [7]

    ReAct: Synergizing reasoning and acting in language models.arXiv preprint arXiv:2210.03629, 2022

    Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik Narasimhan, and Yuan Cao. ReAct: Synergizing reasoning and acting in language models.arXiv preprint arXiv:2210.03629, 2022

    Pith/arXiv arXiv 2022

  8. [8]

    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 Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology, pages 1–22, 2023

    2023

  9. [9]

    A survey on the memory mechanism of large language model-based agents.ACM Transactions on Information Systems, 43(6):1–47, 2025

    Zeyu Zhang, Quanyu Dai, Xiaohe Bo, Chen Ma, Rui Li, Xu Chen, Jieming Zhu, Zhenhua Dong, and Ji-Rong Wen. A survey on the memory mechanism of large language model-based agents.ACM Transactions on Information Systems, 43(6):1–47, 2025

    2025

  10. [10]

    Attentionisallyouneed

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, ŁukaszKaiser,andIlliaPolosukhin. Attentionisallyouneed. InAdvancesinNeuralInformation Processing Systems, pages 5998–6008, 2017

    2017

  11. [11]

    Language models are few-shot learners.Advances in Neural Information Processing Systems, 33:1877–1901, 2020

    TomBBrown,BenjaminMann,NickRyder,MelanieSubbiah,JaredKaplan,PrafullaDhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. Language models are few-shot learners.Advances in Neural Information Processing Systems, 33:1877–1901, 2020

    1901

  12. [12]

    Scaling laws for neural language models.arXiv preprint arXiv:2001.08361, 2020

    Jared Kaplan, Sam McCandlish, Tom Henighan, Tom B Brown, Benjamin Chess, Rewon Child, Scott Gray, Alec Radford, Jeffrey Wu, and Dario Amodei. Scaling laws for neural language models.arXiv preprint arXiv:2001.08361, 2020. 58 From Chatbot to Digital Colleague

    Pith/arXiv arXiv 2001

  13. [13]

    Training compute-optimal large language models

    Jordan Hoffmann, Sebastian Borgeaud, Arthur Mensch, Elena Buchatskaya, Trevor Cai, Eliza Rutherford, Diego de Las Casas, Lisa Anne Hendricks, Johannes Welbl, Aidan Clark, et al. Training compute-optimal large language models. InAdvances in Neural Information Processing Systems (NeurIPS), 2022. URLhttps://arxiv.org/abs/2203.15556

    Pith/arXiv arXiv 2022

  14. [14]

    Gpt-4 technical report.arXiv preprint arXiv:2303.08774, 2023

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

    Pith/arXiv arXiv 2023

  15. [15]

    Chain-of-thought prompting elicits reasoning in large language models

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

    2022

  16. [16]

    Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Peiyi Wang, Qihao Zhu, Runxin Xu, Ruoyu Zhang, Shirong Ma, Xiao Bi, Xiaokang Zhang, Xingkai Yu, Yu Wu, Z. F. Wu, Zhibin Gou, Zhihong Shao, Zhuoshu Li, Ziyi Gao, Aixin Liu, Bing Xue, Bingxuan Wang, Bochao Wu, Bei Feng, Chengda Lu, Chenggang Zhao, Chengqi Deng, Chong Ruan, Damai Dai, Deli Chen, Dongjie Ji, ...

    2025

  17. [17]

    Toolformer: Language models can teach themselves to use tools.Advances in Neural Information Processing Systems, 36, 2023

    Timo Schick, Jane Dwivedi-Yu, Roberto Dessì, Roberta Raileanu, Maria Lomeli, Eric Hambro, Luke Zettlemoyer, Nicola Cancedda, and Thomas Scialom. Toolformer: Language models can teach themselves to use tools.Advances in Neural Information Processing Systems, 36, 2023. 59 From Chatbot to Digital Colleague

    2023

  18. [18]

    ToolLLM: Facilitating large language models to master 16000+ real-world APIs.arXiv preprint arXiv:2307.16789, 2023

    Yujia Qin, Shihao Liang, Yining Ye, Kunlun Zhu, Lan Yan, Yaxi Lu, Yankai Lin, Xin Cong, Xiangru Tang, Bill Qian, et al. ToolLLM: Facilitating large language models to master 16000+ real-world APIs.arXiv preprint arXiv:2307.16789, 2023

    Pith/arXiv arXiv 2023

  19. [19]

    AutoGen: Enabling next-gen LLM applications via multi-agent conversation.arXiv preprint arXiv:2308.08155, 2023

    Qingyun Wu, Gagan Bansal, Jieyu Zhang, Yiran Wu, Beibin Li, Erkang Zhu, Li Jiang, Xiaoyun Zhang, Shaokun Zhang, Jiale Liu, et al. AutoGen: Enabling next-gen LLM applications via multi-agent conversation.arXiv preprint arXiv:2308.08155, 2023

    Pith/arXiv arXiv 2023

  20. [20]

    MetaGPT: Meta programming for multi-agent collaborative framework.arXiv preprint arXiv:2308.00352, 2023

    Sirui Hong, Xiawu Zheng, Jonathan Chen, Yuheng Cheng, Jinlin Wang, Ceyao Zhang, Zili Wang, Steven Ka Shing Yau, Zijuan Lin, Liyang Zhou, et al. MetaGPT: Meta programming for multi-agent collaborative framework.arXiv preprint arXiv:2308.00352, 2023

    Pith/arXiv arXiv 2023

  21. [21]

    Introducing ChatGPT.https://openai.com/blog/chatgpt, 2022

    OpenAI. Introducing ChatGPT.https://openai.com/blog/chatgpt, 2022

    2022

  22. [23]

    Gemma 2: Improving open language models at a practical size, 2024

    Gemma Team, Morgane Riviere, Shreya Pathak, Pier Giuseppe Sessa, Cassidy Hardin, Surya Bhupatiraju, Léonard Hussenot, Thomas Mesnard, Bobak Shahriari, Alexandre Ramé, et al. Gemma 2: Improving open language models at a practical size, 2024

    2024

  23. [24]

    Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de Las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lucile Saulnier, et al

    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. Mistral 7b.arXiv preprint arXiv:2310.06825, 2023. URLhttps://arxiv. org/abs/2310.06825

    Pith/arXiv arXiv 2023

  24. [25]

    Pali-x: On scaling up a multilingual vision and language model.arXiv preprint arXiv:2305.18565, 2023

    Xi Chen, Josip Djolonga, Piotr Padlewski, Basil Mustafa, Soravit Changpinyo, Jialin Wu, Carlos Riquelme Ruiz, Sebastian Goodman, Xiao Wang, Yi Tay, et al. Pali-x: On scaling up a multilingual vision and language model.arXiv preprint arXiv:2305.18565, 2023

    Pith/arXiv arXiv 2023

  25. [26]

    Fabio Petroni, Tim Rocktäschel, Sebastian Riedel, Patrick Lewis, Anton Bakhtin, Yuxiang Wu, and Alexander Miller. Language models as knowledge bases? InProceedings of the 2019 conference on empirical methods in natural language processing and the 9th international joint conference on natural language processing (EMNLP-IJCNLP), pages 2463–2473, 2019

    2019

  26. [27]

    Towards reasoning in large language models: A survey

    Jie Huang and Kevin Chen-Chuan Chang. Towards reasoning in large language models: A survey. InFindingsoftheassociationforcomputationallinguistics: ACL2023, pages1049–1065, 2023

    2023

  27. [28]

    Faith and fate: Limits of transformers on compositionality.Advances in neural information processing systems, 36: 70293–70332, 2023

    Nouha Dziri, Ximing Lu, Melanie Sclar, Xiang Lorraine Li, Liwei Jiang, Bill Yuchen Lin, Sean Welleck, Peter West, Chandra Bhagavatula, Ronan Le Bras, et al. Faith and fate: Limits of transformers on compositionality.Advances in neural information processing systems, 36: 70293–70332, 2023

    2023

  28. [29]

    Large language models still can’t plan (a benchmark for llms on planning and reasoning about change)

    Karthik Valmeekam, Alberto Olmo, Sarath Sreedharan, and Subbarao Kambhampati. Large language models still can’t plan (a benchmark for llms on planning and reasoning about change). InNeurIPS 2022 Foundation Models for Decision Making Workshop, 2022

    2022

  29. [30]

    Scaling LLM test-time compute optimally can be more effective than scaling model parameters, 2024

    Charlie Snell, Jaehoon Lee, Kelvin Xu, and Aviral Kumar. Scaling LLM test-time compute optimally can be more effective than scaling model parameters, 2024. URLhttps:// arxiv.org/abs/2408.03314. 60 From Chatbot to Digital Colleague

    Pith/arXiv arXiv 2024

  30. [31]

    Large language models are zero-shot reasoners.Ad- vances in Neural Information Processing Systems, 35:22199–22213, 2022

    Takeshi Kojima, Shixiang Shane Gu, Machel Reid, Yutaka Matsuo, and Yusuke Iwasawa. Large language models are zero-shot reasoners.Ad- vances in Neural Information Processing Systems, 35:22199–22213, 2022. URL https://proceedings.neurips.cc/paper_files/paper/2022/hash/ 8bb0d291acd4acf06ef112099c16f326-Abstract-Conference.html

    2022

  31. [32]

    Self-refine: Iterative refinement with self- feedback.Advances in Neural Information Processing Systems, 36:46534–46594,

    Aman Madaan, Niket Tandon, Prakhar Gupta, Skyler Hallinan, Luyu Gao, Sarah Wiegreffe, Uri Alon, Nouha Dziri, Shrimai Prabhumoye, Yiming Yang, Shashank Gupta, Bodhisattwa Prasad Majumder, Katherine Hermann, Sean Welleck, Amir Yazdanbakhsh, and Peter Clark. Self-refine: Iterative refinement with self- feedback.Advances in Neural Information Processing Syste...

  32. [33]

    Re- flexion: Language agents with verbal reinforcement learning.Advances in neural information processing systems, 36:8634–8652, 2023

    Noah Shinn, Federico Cassano, Ashwin Gopinath, Karthik Narasimhan, and Shunyu Yao. Re- flexion: Language agents with verbal reinforcement learning.Advances in neural information processing systems, 36:8634–8652, 2023

    2023

  33. [34]

    Let’s verify step by step

    Hunter Lightman, Vineet Kosaraju, Yuri Burda, Harrison Edwards, Bowen Baker, Teddy Lee, Jan Leike, John Schulman, Ilya Sutskever, and Karl Cobbe. Let’s verify step by step. In International Conference on Learning Representations, volume 2024, pages 39578–39601, 2024

    2024

  34. [35]

    Xu, Jun-Mei Song, Mingchuan Zhang, Y

    Zhihong Shao, Peiyi Wang, Qihao Zhu, R. Xu, Jun-Mei Song, Mingchuan Zhang, Y. K. Li, Yu Wu, and Daya Guo. Deepseekmath: Pushing the limits of mathematical reasoning in open language models.ArXiv, abs/2402.03300, 2024

    Pith/arXiv arXiv 2024

  35. [36]

    Generative to agentic ai: Survey, conceptualization, and challenges

    Johannes Schneider. Generative to agentic ai: Survey, conceptualization, and challenges. arXiv preprint arXiv:2504.18875, 2025

    Pith/arXiv arXiv 2025

  36. [37]

    Agentic reasoning for large language models.arXiv preprint arXiv:2601.12538, 2026

    Tianxin Wei, Ting-Wei Li, Zhining Liu, Xuying Ning, Ze Yang, Jiaru Zou, Zhichen Zeng, Ruizhong Qiu, Xiao Lin, Dongqi Fu, et al. Agentic reasoning for large language models.arXiv preprint arXiv:2601.12538, 2026

    Pith/arXiv arXiv 2026

  37. [38]

    Advancing reasoning in large language models: Promising methods and approaches

    Avinash Patil and Aryan Jadon. Advancing reasoning in large language models: Promising methods and approaches. InInternational Conference on Computational Intelligence and Soft Computing, pages 284–298. Springer, 2025

    2025

  38. [39]

    Edge general intelligence through world models, large language models, and agentic ai: Fundamentals, solutions, and challenges

    Changyuan Zhao, Guangyuan Liu, Ruichen Zhang, Yinqiu Liu, Jiacheng Wang, Jiawen Kang, Dusit Niyato, Zan Li, Xuemin Shen, Zhu Han, et al. Edge general intelligence through world models, large language models, and agentic ai: Fundamentals, solutions, and challenges. IEEE Transactions on Cognitive Communications and Networking, 2026

    2026

  39. [40]

    A survey on large language model-based game agents.arXiv preprint arXiv:2404.02039, 2024

    Sihao Hu, Tiansheng Huang, Gaowen Liu, Ramana Rao Kompella, Fatih Ilhan, Selim Furkan Tekin, Yichang Xu, Zachary Yahn, and Ling Liu. A survey on large language model-based game agents.arXiv preprint arXiv:2404.02039, 2024

    Pith/arXiv arXiv 2024

  40. [41]

    Toward large reasoning models: A survey of reinforced reasoning with large language models.Patterns, 6(10), 2025

    Fengli Xu, Qianyue Hao, Chenyang Shao, Zefang Zong, Yu Li, Jingwei Wang, Yunke Zhang, Jingyi Wang, Xiaochong Lan, Jiahui Gong, et al. Toward large reasoning models: A survey of reinforced reasoning with large language models.Patterns, 6(10), 2025. 61 From Chatbot to Digital Colleague

    2025

  41. [42]

    Large language model based multi-agents: A survey of progress and challenges.arXiv preprint arXiv:2402.01680, 2024

    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

    Pith/arXiv arXiv 2024

  42. [43]

    Large language model agents: A comprehensive survey on architectures, capabilities, and applications

    Yiming Lei, Jiawei Xu, Chia Xin Liang, Ziqian Bi, Xiaoming Li, Danyang Zhang, Junhao Song, and Zhenyu Yu. Large language model agents: A comprehensive survey on architectures, capabilities, and applications. 2025

    2025

  43. [44]

    A survey of reasoning with foundation models: Concepts, methodologies, and outlook.ACM Computing Surveys, 57(11):1–43, 2025

    Jiankai Sun, Chuanyang Zheng, Enze Xie, Zhengying Liu, Ruihang Chu, Jianing Qiu, Jiaqi Xu, Mingyu Ding, Hongyang Li, Mengzhe Geng, et al. A survey of reasoning with foundation models: Concepts, methodologies, and outlook.ACM Computing Surveys, 57(11):1–43, 2025

    2025

  44. [45]

    Agentic large language models, a survey.Journal of Artificial Intelligence Research, 84, 2025

    Aske Plaat, Max van Duijn, Niki Van Stein, Mike Preuss, Peter van der Putten, and Kees Joost Batenburg. Agentic large language models, a survey.Journal of Artificial Intelligence Research, 84, 2025

    2025

  45. [46]

    Openhands: An open platform for ai software developers as generalist agents

    Xingyao Wang, Boxuan Li, Yufan Song, Frank F Xu, Xiangru Tang, Mingchen Zhuge, Jiayi Pan, Yueqi Song, Bowen Li, Jaskirat Singh, et al. Openhands: An open platform for ai software developers as generalist agents. InInternational Conference on Learning Representations, volume 2025, pages 65882–65919, 2025

    2025

  46. [47]

    SWE-agent: Agent-computer interfaces enable automated software engineering

    John Yang, Carlos E Jimenez, Alexander Wettig, Kilian Lieret, Shunyu Yao, Karthik R Narasimhan, and Ofir Press. SWE-agent: Agent-computer interfaces enable automated software engineering. InThe Thirty-eighth Annual Conference on Neural Information Process- ing Systems, 2024. URLhttps://openreview.net/forum?id=mXpq6ut8J3

    2024

  47. [48]

    Workarena: How capable are web agents at solving common knowledge work tasks?arXiv preprint arXiv:2403.07718, 2024

    Alexandre Drouin, Maxime Gasse, Massimo Caccia, Issam H Laradji, Manuel Del Verme, Tom Marty, Léo Boisvert, Megh Thakkar, Quentin Cappart, David Vazquez, et al. Workarena: How capable are web agents at solving common knowledge work tasks?arXiv preprint arXiv:2403.07718, 2024

    Pith/arXiv arXiv 2024

  48. [49]

    OSWorld: Benchmarking multimodal agents for open-ended tasks in real computer environments

    Tianbao Xie, Danyang Zhang, Jixuan Chen, Xiaochuan Li, Siheng Zhao, Ruisheng Cao, Toh Jing Hua, Zhoujun Cheng, Dongchan Shin, Fangyu Lei, Yitao Liu, Yiheng Xu, Shuyan Zhou, Silvio Savarese, Caiming Xiong, Victor Zhong, and Tao Yu. OSWorld: Benchmarking multimodal agents for open-ended tasks in real computer environments. InThe Thirty-eight Conference on N...

    2024

  49. [50]

    Xu, Hao Zhu, Xuhui Zhou, Robert Lo, Abishek Sridhar, Xianyi Cheng, Tianyue Ou, Yonatan Bisk, Daniel Fried, Uri Alon, and Graham Neubig

    Shuyan Zhou, Frank F. Xu, Hao Zhu, Xuhui Zhou, Robert Lo, Abishek Sridhar, Xianyi Cheng, Tianyue Ou, Yonatan Bisk, Daniel Fried, Uri Alon, and Graham Neubig. Webarena: A realistic web environment for building autonomous agents. InThe Twelfth International Conference on Learning Representations, 2024. URLhttps://openreview.net/forum? id=oKn9c6ytLx

    2024

  50. [51]

    Voyager: An open-ended embodied agent with large language models.Transactions on Machine Learning Research, 2024

    Guanzhi Wang, Yuqi Xie, Yunfan Jiang, Ajay Mandlekar, Chaowei Xiao, Yuke Zhu, Linxi Fan, and Anima Anandkumar. Voyager: An open-ended embodied agent with large language models.Transactions on Machine Learning Research, 2024. ISSN 2835-8856. URLhttps: //openreview.net/forum?id=ehfRiF0R3a. 62 From Chatbot to Digital Colleague

    2024

  51. [52]

    Large language model agent: A survey on methodology, applications and challenges.arXiv preprint arXiv:2503.21460, 2025

    Junyu Luo, Weizhi Zhang, Ye Yuan, Yusheng Zhao, Junwei Yang, Yiyang Gu, Bohan Wu, Binqi Chen, Ziyue Qiao, Qingqing Long, et al. Large language model agent: A survey on methodology, applications and challenges.arXiv preprint arXiv:2503.21460, 2025

    Pith/arXiv arXiv 2025

  52. [53]

    Beyondgenerativeartificial intelligence: Roadmap for natural language generation.arXiv preprint arXiv:2407.10554, 2024

    María Miró Maestre, Iván Martínez-Murillo, Tania J Martin, Borja Navarro-Colorado, Antonio Ferrández, Armando Suárez Cueto, and Elena Lloret. Beyondgenerativeartificial intelligence: Roadmap for natural language generation.arXiv preprint arXiv:2407.10554, 2024

    Pith/arXiv arXiv 2024

  53. [54]

    A survey on the optimization of large language model-based agents.ACM Computing Surveys, 58(9):1–37, 2026

    Shangheng Du, Jiabao Zhao, Jinxin Shi, Zhentao Xie, Xin Jiang, Yanhong Bai, and Liang He. A survey on the optimization of large language model-based agents.ACM Computing Surveys, 58(9):1–37, 2026

    2026

  54. [55]

    Exploring autonomous agents through the lens of large language models: A review.arXiv preprint arXiv:2404.04442, 2024

    Saikat Barua. Exploring autonomous agents through the lens of large language models: A review.arXiv preprint arXiv:2404.04442, 2024

    Pith/arXiv arXiv 2024

  55. [56]

    A survey on agentic multimodal large language models.arXiv preprint arXiv:2510.10991, 2025

    Huanjin Yao, Ruifei Zhang, Jiaxing Huang, Jingyi Zhang, Yibo Wang, Bo Fang, Ruolin Zhu, Yongcheng Jing, Shunyu Liu, Guanbin Li, et al. A survey on agentic multimodal large language models.arXiv preprint arXiv:2510.10991, 2025

    arXiv 2025

  56. [57]

    AgentBench: Evaluating LLMs as agents.arXiv preprint arXiv:2308.03688, 2023

    Xiao Liu, Hao Yu, Hanchen Zhang, Yifan Xu, Xuanyu Lei, Hanyu Lai, Yu Gu, Hangliang Ding, Kaiwen Men, Kejuan Yang, et al. AgentBench: Evaluating LLMs as agents.arXiv preprint arXiv:2308.03688, 2023

    Pith/arXiv arXiv 2023

  57. [58]

    Jimenez et al

    Carlos E. Jimenez et al. Swe-bench: Can language models resolve real-world github issues? arXiv preprint arXiv:2310.06770, 2023. doi: 10.48550/arXiv.2310.06770. URLhttps: //arxiv.org/abs/2310.06770

    Pith/arXiv arXiv doi:10.48550/arxiv.2310.06770 2023

  58. [59]

    A survey on hallucination in large language models: Principles, taxonomy, challenges, and open questions.arXiv preprint arXiv:2311.05232, 2023

    Lei Huang, Weijiang Yu, Weitao Ma, Weihong Zhong, Zhangyin Feng, Haotian Wang, Qian- glong Chen, Weihua Peng, Xiaocheng Feng, Bing Qin, et al. A survey on hallucination in large language models: Principles, taxonomy, challenges, and open questions.arXiv preprint arXiv:2311.05232, 2023

    Pith/arXiv arXiv 2023

  59. [60]

    Survey of hallucination in natural language generation

    Ziwei Ji, Nayeon Lee, Rita Frieske, Tiezheng Yu, Dan Su, Yan Xu, Etsuko Ishii, Ye Jin Bang, Andrea Madotto, and Pascale Fung. Survey of hallucination in natural language generation. ACM Computing Surveys, 55(12):1–38, 2023

    2023

  60. [61]

    Memgpt: towards llms as operating systems

    Charles Packer, Vivian Fang, Shishir_G Patil, Kevin Lin, Sarah Wooders, and Joseph_E Gonzalez. Memgpt: towards llms as operating systems. 2023

    2023

  61. [62]

    Identifying the risks of lm agents with an lm-emulated sandbox

    Yangjun Ruan, Honghua Dong, Andrew Wang, Silviu Pitis, Yongchao Zhou, Jimmy Ba, Yann Dubois, Chris Maddison, and Tatsunori Hashimoto. Identifying the risks of lm agents with an lm-emulated sandbox. InInternational Conference on Learning Representations, volume 2024, pages 27031–27098, 2024

    2024

  62. [63]

    Agentdojo: A dynamic environment to evaluate prompt injection attacks and defenses for llm agents.Advances in Neural Information Processing Systems, 37:82895–82920, 2024

    Edoardo Debenedetti, Jie Zhang, Mislav Balunovic, Luca Beurer-Kellner, Marc Fischer, and Florian Tramèr. Agentdojo: A dynamic environment to evaluate prompt injection attacks and defenses for llm agents.Advances in Neural Information Processing Systems, 37:82895–82920, 2024

    2024

  63. [64]

    Injecagent: Benchmarking indirect prompt injections in tool-integrated large language model agents

    Qiusi Zhan, Zhixiang Liang, Zifan Ying, and Daniel Kang. Injecagent: Benchmarking indirect prompt injections in tool-integrated large language model agents. InFindings of the Association for Computational Linguistics: ACL 2024, pages 10471–10506, 2024. 63 From Chatbot to Digital Colleague

    2024

  64. [65]

    Openclaw prism: A zero-fork, defense-in-depth runtime security layer for tool- augmented llm agents.arXiv preprint arXiv:2603.11853, 2026

    Frank Li. Openclaw prism: A zero-fork, defense-in-depth runtime security layer for tool- augmented llm agents.arXiv preprint arXiv:2603.11853, 2026. URL https://arxiv. org/abs/2603.11853

    arXiv 2026

  65. [67]

    Chameleon: Plug-and-play compositional reasoning with large language models.Advances in Neural Information Processing Systems, 36:43447–43478, 2023

    Pan Lu, Baolin Peng, Hao Cheng, Michel Galley, Kai-Wei Chang, Ying Nian Wu, Song-Chun Zhu, and Jianfeng Gao. Chameleon: Plug-and-play compositional reasoning with large language models.Advances in Neural Information Processing Systems, 36:43447–43478, 2023

    2023

  66. [68]

    Qwen3.6-27B non-thinking, April 2026

    Qwen Team. Qwen3.6-27B non-thinking, April 2026. URLhttps://qwen.ai/blog?id= qwen3.6-27b

    2026

  67. [69]

    Stateless yet not forgetful: Implicit memory as a hidden channel in llms.arXiv preprint arXiv:2602.08563, 2026

    Ahmed Salem, Andrew Paverd, and Sahar Abdelnabi. Stateless yet not forgetful: Implicit memory as a hidden channel in llms.arXiv preprint arXiv:2602.08563, 2026

    arXiv 2026

  68. [70]

    Reasoning and planning with large language models in code development

    Hao Ding, Ziwei Fan, Ingo Guehring, Gaurav Gupta, Wooseok Ha, Jun Huan, Linbo Liu, Behrooz Omidvar-Tehrani, Shiqi Wang, and Hao Zhou. Reasoning and planning with large language models in code development. InProceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 6480–6490, 2024

    2024

  69. [71]

    Structured chain-of-thought prompting for code generation.ACM Transactions on Software Engineering and Methodology, 34(2):1–23, 2025

    Jia Li, Ge Li, Yongmin Li, and Zhi Jin. Structured chain-of-thought prompting for code generation.ACM Transactions on Software Engineering and Methodology, 34(2):1–23, 2025

    2025

  70. [72]

    A survey on large reasoning models with self-play deep reinforcement learning and chain-of-thought

    Chunfang Ou, Lijuan Fan, Guobin Fu, Renzheng Liu, and Zhongzhi Li. A survey on large reasoning models with self-play deep reinforcement learning and chain-of-thought. In Proceedings of the 2025 2nd Symposium on Big Data, Neural Networks, and Deep Learning, pages 186–190, 2025

    2025

  71. [73]

    Mul- timodal chain-of-thought reasoning in language models.arXiv preprint arXiv:2302.00923, 2023

    Zhuosheng Zhang, Aston Zhang, Mu Li, Hai Zhao, George Karypis, and Alex Smola. Mul- timodal chain-of-thought reasoning in language models.arXiv preprint arXiv:2302.00923, 2023

    Pith/arXiv arXiv 2023

  72. [74]

    Progco: Program helps self-correction of large language models

    Xiaoshuai Song, Yanan Wu, Weixun Wang, Jiaheng Liu, Wenbo Su, and Bo Zheng. Progco: Program helps self-correction of large language models. InProceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 944–959, 2025

    2025

  73. [75]

    Thinking, fast and slow.Farrar, Straus and Giroux, 2011

    Daniel Kahneman. Thinking, fast and slow.Farrar, Straus and Giroux, 2011

    2011

  74. [76]

    A survey of large language models.arXiv preprint arXiv:2303.18223, 2023

    Wayne Xin Zhao, Kun Zhou, Junyi Li, Tianyi Tang, Xia Wang, Yupeng Hou, Yingqian Min, Beichen Zhang, Junjie Zhang, Zican Dong, et al. A survey of large language models.arXiv preprint arXiv:2303.18223, 2023

    Pith/arXiv arXiv 2023

  75. [77]

    Safechain: Safety of language models with long chain-of-thought reasoning capabilities

    Fengqing Jiang, Zhangchen Xu, Yuetai Li, Luyao Niu, Zhen Xiang, Bo Li, Bill Yuchen Lin, and Radha Poovendran. Safechain: Safety of language models with long chain-of-thought reasoning capabilities. InFindings of the Association for Computational Linguistics: ACL 2025, pages 23303–23320, 2025. 64 From Chatbot to Digital Colleague

    2025

  76. [78]

    Yancheng He, Shilong Li, Jiaheng Liu, Weixun Wang, Xingyuan Bu, Ge Zhang, Zy Peng, Zhaoxiang Zhang, Zhicheng Zheng, Wenbo Su, et al. Can large language models detect errors in long chain-of-thought reasoning? InProceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 18468–18489, 2025

    2025

  77. [79]

    Interactive reasoning: Visualizing and controlling chain- of-thought reasoning in large language models

    Rock Yuren Pang, KJ Kevin Feng, Shangbin Feng, Chu Li, Weijia Shi, Yulia Tsvetkov, Jeffrey Heer, and Katharina Reinecke. Interactive reasoning: Visualizing and controlling chain- of-thought reasoning in large language models. InProceedings of the 31st International Conference on Intelligent User Interfaces, pages 852–867, 2026

    2026

  78. [80]

    Chain-of-thought matters: improving long- context language models with reasoning path supervision.arXiv preprint arXiv:2502.20790, 2025

    Dawei Zhu, Xiyu Wei, Guangxiang Zhao, Wenhao Wu, Haosheng Zou, Junfeng Ran, Xun Wang, Lin Sun, Xiangzheng Zhang, and Sujian Li. Chain-of-thought matters: improving long- context language models with reasoning path supervision.arXiv preprint arXiv:2502.20790, 2025

    Pith/arXiv arXiv 2025

  79. [81]

    Aligning large and small language models via chain- of-thought reasoning

    Leonardo Ranaldi and Andre Freitas. Aligning large and small language models via chain- of-thought reasoning. InProceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1812–1827, 2024

    2024

  80. [82]

    Recot: Reflective self- correction training for mitigating confirmation bias in large vision-language models

    Mengxue Qu, Yibo Hu, Kunyang Han, Yunchao Wei, and Yao Zhao. Recot: Reflective self- correction training for mitigating confirmation bias in large vision-language models. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 9147–9157, 2025

    2025

Showing first 80 references.