Pith. sign in

REVIEW 4 major objections 5 minor 4 cited by

PKRD-CoT: A Unified Chain-of-thought Prompting for Multi-Modal Large Language Models in Autonomous Driving

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read A chain-of-thought prompt built on perception, knowledge, reasoning, and decision-making lifts GPT-4.0's driving decision accuracy to 94 percent on 100 real scenes, from 72 percent with zero-shot prompting.

desk verdict A sensible zero-shot CoT prompt for MLLM driving agents with a plausible 22-point accuracy gain over plain zero-shot, but the evaluation is too thin and under-specified to accept the paper's stronger claims yet. read the letter →

arxiv 2412.02025 v1 pith:4FBRS65U submitted 2024-12-02 cs.RO cs.AI

Xuewen Luo , Fan Ding , Yinsheng Song , Xiaofeng Zhang , Junnyong Loo This is my paper · ORCID
classification cs.ROcs.AI
keywords zero-shotchain-of-thoughtmultimodallargelanguagemodelsautonomousdrivingpromptengineeringPKRD-CoTknowledge-drivendecision-making
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 claims that a zero-shot chain-of-thought prompt structured around the four driving capabilities of perception, knowledge, reasoning, and decision-making lets multimodal large language models make more accurate driving decisions without any fine-tuning. On 100 real driving scenes selected from the NuScenes dataset, GPT-4.0 with this prompt, named PKRD-CoT, reached 94 percent decision accuracy, versus 72 percent for plain zero-shot prompting and 88 percent for role-playing prompting. The authors present this as a low-cost alternative to training end-to-end autonomous driving models, because it works by prompting an existing model rather than training a new one. They also use the same prompt to compare six MLLMs on perception, knowledge, mathematical reasoning, and decision-making, with GPT-4.0 performing best overall.

What carries the argument

PKRD-CoT is a prompt template that forces the model to proceed through four named steps: Observation (describe the current environment), Identification (detect and localize relevant objects), Memory (store the scene understanding in a structured JSON record), and Decision (choose a driving action from options such as speed up, slow down, stop, maintain, or change lane). The chain-of-thought structure is the load-bearing mechanism: it makes the model's reasoning explicit and interpretable, and the JSON memory module is intended to compensate for the model's limited context over prolonged interactions.

What would settle it

Run the same PKRD-CoT template on a larger, independently annotated set of driving scenes with a predefined scoring rubric and compare accuracy against plain zero-shot prompting. If the gap shrinks to within noise or reverses, the claimed benefit does not hold. A second check would be to run the exact same prompt template and scoring procedure with an open-weight MLLM to rule out effects specific to GPT-4.0.

Watch

Extended reading notes

Core claim

The central claim is that decomposing a driving task into Perception, Knowledge, Reasoning, and Decision-making steps -- with each step prompted explicitly and intermediate scene understanding written to a JSON memory -- produces driving decisions that are more accurate and more interpretable than decisions from generic zero-shot or role-playing prompts. The paper's main demonstration is the GPT-4.0 ablation on 100 real scene samples, where the structured prompt scores 94 percent against 72 percent for zero-shot and 88 percent for role-playing. The same PKRD-CoT template is then applied to five other MLLMs to produce a comparative evaluation across the four capability axes. The intended upshot is that knowledge-driven, prompt-based driving can substitute for expensive training pipelines.

Load-bearing premise

The claim rests on the assumption that 'correct' driving decisions are well-defined for the 100 selected real scenes and that those scenes represent the range of real driving situations; the paper does not state the sample selection criteria, the labeling protocol, or confidence intervals, so the measured gap could be shaped by sample choice and subjective scoring.

Editorial extensions

If this is right

  • MLLMs become usable as driving decision agents without retraining, lowering the cost and data requirements for autonomous driving development.
  • A single prompt template transfers across different MLLMs, so upgrades to the base models can translate directly into improved driving decisions.
  • The structured JSON output from the memory step could be fed into downstream planning or control modules as interpretable state information.
  • The four-capability evaluation protocol offers a reusable way to assess whether an MLLM is suitable for autonomous driving tasks before deployment.

Reading between the lines

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

  • The 94 percent figure rests on only 100 scenes with no stated selection or annotation protocol; an independent, larger-scale labeled evaluation would be needed to confirm that the accuracy gap over zero-shot prompting is real rather than an artifact of sample choice.
  • If the effect is genuine, the same observation-knowledge-reasoning-decision decomposition could be applied to other sequential high-stakes domains, such as remote robot operation or clinical decision support.
  • Combining PKRD-CoT with a conventional planner or safety filter could yield a hybrid system where the MLLM proposes high-level actions and the safety layer vetoes unsafe ones, addressing the paper's noted safety concern.
  • Testing with an open-weight model whose weights and prompt template are frozen would clarify how much of the gain comes from the prompt structure versus the specific commercial model.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes PKRD-CoT, a zero-shot chain-of-thought prompt framework for multimodal large language models (MLLMs) in autonomous driving. The framework decomposes the driving task into four stages—perception, knowledge, reasoning, and decision-making—and is evaluated against zero-shot and role-playing baselines on 100 real-scene samples from NuScenes, reporting decision accuracies of 72%, 88%, and 94% for GPT-4.0. The paper also benchmarks six MLLMs (GPT-4.0, Claude, LLava1.6, Qwen-VL-Plus, CogVLM, and Minigpt4) across perception, knowledge, mathematical reasoning, and decision-making, concluding that GPT-4.0 performs best overall.

Significance. The idea of a modular, prompt-only decomposition of driving decisions is timely and, if validated, would offer a low-cost alternative to fine-tuning MLLMs for autonomous driving. The multi-model comparison on capability-specific tasks is a useful contribution to the community. However, the central claim of improved decision accuracy rests on a small, undocumented evaluation, and the qualitative sections do not yet support the strong conclusions drawn. The framework is promising, but the evidence base needs substantial strengthening before the results can be considered established.

major comments (4)
  1. [Section 3.3, Table 1] The headline claim that PKRD-CoT improves decision accuracy is supported only by 100 samples without error bars, significance tests, sample-selection criteria, or a documented annotation rubric. At n=100, the 94% versus 88% difference over role-playing is within sampling noise: a simple two-proportion standard error gives a 95% confidence interval for the difference of roughly -0.02 to +0.14. The 22-point gap over zero-shot is larger, but without knowing how the 100 scenes were chosen and who decided what counts as a 'correct decision', both gaps could reflect sample selection and subjective scoring. The manuscript must provide the annotation protocol, inter-annotator agreement, and ideally a larger or independent evaluation set.
  2. [Sections 3.2 and 3.3] The exact PKRD-CoT prompt is never shown. Since the contribution is a prompt-design framework, the full template—including the wording of the perception, knowledge, reasoning, and decision steps and the JSON memory format—is essential for reproducibility and for ruling out confounds such as prompt length or task specificity.
  3. [Sections 4.1-4.4, Tables 2-3] The benchmark comparisons are not statistically grounded. Tables 2 and 3 do not report sample sizes; Table 3 reports 'Pythagorean Theorem Correct' rates without stating how many distance-calculation trials were used, and Minigpt4's '-' entry is only explained narratively. Sections 4.2 and 4.4 evaluate knowledge and decision-making through qualitative visual examples rather than quantitative metrics, so the rankings of models in the conclusion are not supported by the evidence presented.
  4. [Sections 3.1 and 4] There is a potential circularity: PKRD-CoT is defined by the four capabilities of perception, knowledge, reasoning, and decision-making, and the evaluation framework assesses the same four capabilities. This does not invalidate the framework, but it means the benchmark is tailored to its own decomposition. An external validation using a driving-specific ground truth—such as closed-loop simulation metrics or agreement with human expert decisions—is needed to test whether the four-step decomposition actually improves driving quality.
minor comments (5)
  1. [Throughout] There are multiple typos and grammatical errors, e.g., 'There are growing interest' in the abstract, 'through process' in Section 1, 'the the understanding' in Section 1, and 'chain-of-mind cues' in the conclusion.
  2. [References] Several references do not match the cited models: reference [18] cites Koubaa's GPT-4 versus GPT-3.5 comparison for GPT-4.0, reference [25] cites Touvron et al.'s Llama 2 for LLava1.6, and reference [9] is incomplete ('Contributors: Drivelm: Drive on language').
  3. [Section 3.3] The prompt name is spelled 'PRKD-CoT' in one sentence; please make the acronym consistent throughout.
  4. [Table 3] The column header 'Pythagorean Theorem Correct' should clarify what the checkmark and cross mean, and the number of trials used to compute each percentage should be stated.
  5. [Section 4.1] The text says a model gets 100% or 0% per target, but the reported averages (e.g., 90%, 77.78%) imply different numbers of samples per category; please report the sample counts.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper reports empirical prompt evaluations without self-citations or fitted parameters, and its claims do not reduce to their inputs.

full rationale

PKRD-CoT is an empirical prompting study, not a derivation. The central quantitative claim in Table 1 (94% vs 72% vs 88% decision accuracy on 100 NuScenes samples) is a measured outcome, not a quantity constructed from the prompt's definitions. The paper contains no fitted parameters, no equations whose outputs equal an input, and no load-bearing self-citations; the reference list does not include prior works by the present authors. The evaluation dimensions in Section 4 intentionally mirror the prompt's four capabilities, and the decision-accuracy rubric is not fully documented, which are legitimate reproducibility and validity concerns; however, they do not make the reported result equivalent to the framework's definition by construction. The absence of confidence intervals and the fact that the 6% gain over role-playing is within sampling noise are statistical-reporting weaknesses, not circularity.

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

The paper introduces no new free parameters, equations, or physical entities. Its load-bearing assumptions are about the sufficiency of the four-step capability decomposition, the transferability of zero-shot CoT to driving, and the reliability of the authors' subjective correctness labels.

assumptions (3)
  • ad hoc to paper The four capabilities (perception, knowledge, reasoning, decision-making) are a sufficient and appropriate decomposition for autonomous driving decision quality.
    Section 3.1 introduces the framework as derived from these four capabilities, but no independent evidence shows that this decomposition captures the factors that drive decision accuracy.
  • domain assumption Zero-shot CoT prompting can improve MLLM reasoning in driving without task-specific training.
    Borrowed from prior CoT literature (refs 15-16) and assumed to transfer to multimodal driving, without validation on this domain beyond the paper's own small experiments.
  • domain assumption A human-written 'correct decision' label for each image is a reliable ground truth for driving quality.
    Section 3.3 defines accuracy as correct decisions divided by total samples, but the labeling process, criteria, and inter-annotator agreement are not described.

how reviews work

0 comments
Cite this review

Pith. "Pith review of PKRD-CoT: A Unified Chain-of-thought Prompting for Multi-Modal Large Language Models in Autonomous Driving." pith.science (2026). https://pith.science/paper/4FBRS65U

@misc{pith2026241202025,
  author       = {Pith},
  title        = {Pith review of: PKRD-CoT: A Unified Chain-of-thought Prompting for Multi-Modal Large Language Models in Autonomous Driving},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4FBRS65U}},
  note         = {Machine review of arXiv:2412.02025}
}
read the original abstract

There is growing interest in leveraging the capabilities of robust Multi-Modal Large Language Models (MLLMs) directly within autonomous driving contexts. However, the high costs and complexity of designing and training end-to-end autonomous driving models make them challenging for many enterprises and research entities. To address this, our study explores a seamless integration of MLLMs into autonomous driving systems by proposing a Zero-Shot Chain-of-Thought (Zero-Shot-CoT) prompt design named PKRD-CoT. PKRD-CoT is based on the four fundamental capabilities of autonomous driving: perception, knowledge, reasoning, and decision-making. This makes it particularly suitable for understanding and responding to dynamic driving environments by mimicking human thought processes step by step, thus enhancing decision-making in real-time scenarios. Our design enables MLLMs to tackle problems without prior experience, thereby increasing their utility within unstructured autonomous driving environments. In experiments, we demonstrate the exceptional performance of GPT-4.0 with PKRD-CoT across autonomous driving tasks, highlighting its effectiveness in autonomous driving scenarios. Additionally, our benchmark analysis reveals the promising viability of PKRD-CoT for other MLLMs, such as Claude, LLava1.6, and Qwen-VL-Plus. Overall, this study contributes a novel and unified prompt-design framework for GPT-4.0 and other MLLMs in autonomous driving, while also rigorously evaluating the efficacy of these widely recognized MLLMs in the autonomous driving domain through comprehensive comparisons.

Figures

Figures reproduced from arXiv: 2412.02025 by the authors.

Figure 1
Figure 1. The MLLM model operates as a driver agent within the PKRD-CoT paradigm, including a dynamic environment, an MLLM model with capabilities in perception, knowledge, reasoning, and decision-making, and a memory module that stores infor￾mation in JSON format. The MLLM model continuously senses the environment, rec￾ognizes targets, reasons about situations, and interacts with the memory module to make decisions for contr… view at source ↗
Figure 2
Figure 2. Example Outputs of GPT4.0 with PKRD-CoT in Autonomous Driving In our experiments, we tested our proposed PKRD-CoT on three scenarios, which includes highway, daytime traffic and nighttime traffic, focusing on cars, pedestrians, traffic signs (traffic lights and zebra crosswalks), and the environ￾ment around the car (daytime, nighttime, or rainy or sunny days). For object detection in image space of the cars, pedestr… view at source ↗
Figure 3
Figure 3. Example of Ablation Experiment Outputs Current Scene. The evaluation criteria are based on the model’s ability to accu￾rately identify the target species and generate the corresponding descriptions: if the model correctly recognizes and describes the target, the perceptual accuracy for that target species is taken to be 100%; ; conversely, if the recognition is incorrect, the perceptual accuracy is 0%. The average p… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: CogVLM chat: Positioning of Car and Pedestrian From the quantitative results in [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]
Figure 5
Figure 5. Figure 5: Visual Results of Knowledge Ability Comparative Experiments derstanding and adhering to traffic rules. On the contrary, Minigpt4 is inclined towards utilizing information of upcoming signal changes for making decisions, thus showing an ability to predict the environmen…
Figure 6
Figure 6. Figure 6: Visual Results of Decision-Making Ability Comparative Experiments Based on the results, GPT-4.0, Claude, and LLava1.6 adopt a cautious ap￾proach to decision-making, preferring to maintain the current state until a signif￾icant need for change arises. This method is eff…

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. Chain-of-Thought for Autonomous Driving: A Comprehensive Survey and Future Prospects

    cs.RO 2025-05 conditional novelty 4.0 of 10

    A survey that classifies chain-of-thought methods for autonomous driving into modular, logical, and reflective pipelines, and proposes three evolutionary stages from direct prompting to reinforcement learning.

  2. X-Driver: Explainable Autonomous Driving with Vision-Language Models

    cs.RO 2025-05 conditional novelty 4.0 of 10

    X-Driver combines LLaVA with chain-of-thought prompts to produce explainable driving decisions, reporting modest gains over UniAD on the Bench2Drive closed-loop benchmark.

  3. SenseRAG: Constructing Environmental Knowledge Bases with Proactive Querying for LLM-Based Autonomous Driving

    cs.AI 2025-01 reject novelty 4.0 of 10

    A retrieval-augmented LLM framework that lets a driving model query a database of environmental sensor data reduces reported trajectory prediction error by roughly 70 percent, but the evaluation design inflates the gain.

  4. Evaluation of LLMs for mathematical problem solving

    cs.AI 2025-05 reject novelty 3.0 of 10

    A three-model, three-dataset LLM math evaluation using a multi-dimensional reasoning rubric, undermined by contradictory accuracy tables.

Reference graph

Works this paper leans on

40 extracted references · 17 canonical work pages · cited by 4 Pith papers

  1. [1]

    Anthropic (2024), online

    Anthropic: The claude 3 model family: Opus, sonnet, haiku. Anthropic (2024), online

    work page 2024

  2. [2]

    arXiv preprint arXiv:2308.12966 (2023)

    Bai, J., et al.: Qwen-vl: A frontier large vision-language model with versatile abil- ities. arXiv preprint arXiv:2308.12966 (2023)

    arXiv 2023

  3. [3]

    In: Proceedings of the IEEE/CVF International Conference on Com- puter Vision

    Bogdoll, D., et al.: Description of corner cases in automated driving: Goals and challenges. In: Proceedings of the IEEE/CVF International Conference on Com- puter Vision. pp. 1023–1028 (2021)

    work page 2021

  4. [4]

    In: 2019 IEEE Intelligent vehicles symposium (IV)

    Bolte, J.A., et al.: Towards corner case detection for autonomous driving. In: 2019 IEEE Intelligent vehicles symposium (IV). pp. 438–445 (2019)

    work page 2019

  5. [5]

    In: Advances in neural information processing systems

    Brown, T., et al.: Language models are few-shot learners. In: Advances in neural information processing systems. vol. 33, pp. 1877–1901 (2020)

    work page 2020

  6. [6]

    In: Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion

    Caesar, H., et al.: nuscenes: A multimodal dataset for autonomous driving. In: Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion. pp. 11621–11631 (2020)

    work page 2020

  7. [7]

    arXiv preprint arXiv:2306.16927 (2023)

    Chen, L., et al.: End-to-end autonomous driving: Challenges and frontiers. arXiv preprint arXiv:2306.16927 (2023)

    arXiv 2023

  8. [8]

    arXiv preprint arXiv:2310.01957 (2023)

    Chen, L., et al.: Driving with llms: Fusing object-level vector modality for explain- able autonomous driving. arXiv preprint arXiv:2310.01957 (2023)

    arXiv 2023

Show all 40 references
  1. [9]

    Contributors: Drivelm: Drive on language (2023)

    2023

  2. [10]

    In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision

    Cui, C., et al.: A survey on multimodal large language models for autonomous driving. In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. pp. 958–979 (2024)

    2024

  3. [11]

    Advances in Neural Information Processing Systems36 (2024)

    Dai, W., et al.: Instructblip: Towards general-purpose vision-language models with instruction tuning. Advances in Neural Information Processing Systems36 (2024)

    2024

  4. [12]

    arXiv preprint arXiv:2012.15723 (2020)

    Gao, T., et al.: Making pre-trained language models better few-shot learners. arXiv preprint arXiv:2012.15723 (2020)

    2020 arXiv

  5. [13]

    arXiv preprint arXiv:2106.09685 (2021)

    Hu, E., et al.: Lora: Low-rank adaptation of large language models. arXiv preprint arXiv:2106.09685 (2021)

    2021 arXiv

  6. [14]

    arXiv preprint arXiv:2309.12289 (2023)

    Kochdumper, N., Bak, S.: Real-time capable decision making for autonomous driv- ing using reachable sets. arXiv preprint arXiv:2309.12289 (2023)

    2023 arXiv

  7. [15]

    arXiv preprint arXiv:2205.11916 (2022)

    Kojima, T., et al.: Large language models are zero-shot reasoners. arXiv preprint arXiv:2205.11916 (2022)

    2022 arXiv

  8. [16]

    In: Advances in neural information processing systems

    Kojima, T., et al.: Large language models are zero-shot reasoners. In: Advances in neural information processing systems. vol. 35, pp. 22199–22213 (2022)

    2022

  9. [17]

    arXiv preprint arXiv:2308.07702 (2023)

    Kong, A., et al.: Better zero-shot reasoning with role-play prompting. arXiv preprint arXiv:2308.07702 (2023)

    2023 arXiv

  10. [18]

    gpt-3.5: A concise showdown (2023) Title Suppressed Due to Excessive Length 15

    Koubaa, A.: Gpt-4 vs. gpt-3.5: A concise showdown (2023) Title Suppressed Due to Excessive Length 15

    2023

  11. [19]

    In: International Conference on Neural Information Processing

    Li, B., Li, F., Gao, S., Fan, Q., Lu, Y., Hu, R., Zhao, Z.: Efficient prompt tuning for vision and language models. In: International Conference on Neural Information Processing. pp. 77–89. Springer (2023)

    2023

  12. [20]

    IEEE Transactions on Intelligent Transportation Systems (2020)

    Li, P., et al.: A survey on deep learning-based approaches for autonomous driving. IEEE Transactions on Intelligent Transportation Systems (2020)

    2020

  13. [21]

    arXiv preprint arXiv:2312.04316 (2023)

    Li, X., et al.: Towards knowledge-driven autonomous driving. arXiv preprint arXiv:2312.04316 (2023)

    2023 arXiv

  14. [22]

    arXiv preprint arXiv:2107.13586 (2021)

    Liu, P., et al.: Pre-train, prompt, and predict: A systematic survey of prompting methods in natural language processing. arXiv preprint arXiv:2107.13586 (2021)

    2021 arXiv

  15. [23]

    arXiv preprint arXiv:2310.14799 (2023)

    Qin, L., et al.: Cross-lingual prompting: Improving zero-shot chain-of-thought rea- soning across languages. arXiv preprint arXiv:2310.14799 (2023)

    2023 arXiv

  16. [24]

    arXiv preprint arXiv:2312.07488 (2023)

    Shao, H., et al.: Lmdrive: Closed-loop end-to-end driving with large language mod- els. arXiv preprint arXiv:2312.07488 (2023)

    2023 arXiv

  17. [25]

    arXiv preprint arXiv:2307.09288 (2023)

    Touvron, H., et al.: Llama 2: Open foundation and fine-tuned chat models. arXiv preprint arXiv:2307.09288 (2023)

    2023 arXiv

  18. [26]

    arXiv preprint arXiv:2403.15886 (2024)

    Vöge, L., et al.: Leveraging zero-shot prompting for efficient language model dis- tillation. arXiv preprint arXiv:2403.15886 (2024)

    2024 arXiv

  19. [27]

    arXiv preprint arXiv:2305.04091 (2023)

    Wang, L., et al.: Plan-and-solve prompting: Improving zero-shot chain-of-thought reasoning by large language models. arXiv preprint arXiv:2305.04091 (2023)

    2023 arXiv

  20. [28]

    arXiv preprint arXiv:2310.10357 (2023)

    Wang, P., et al.: Bevgpt: Generative pre-trained large model for autonomous driv- ing prediction, decision-making, and planning. arXiv preprint arXiv:2310.10357 (2023)

    2023 arXiv

  21. [29]

    arXiv preprint arXiv:2310.17642 (2023)

    Wang, T.H., et al.: Drive anywhere: Generalizable end-to-end autonomous driving with multi-modal foundation models. arXiv preprint arXiv:2310.17642 (2023)

    2023 arXiv

  22. [30]

    arXiv preprint arXiv:2311.03079 (2023)

    Wang, W., et al.: Cogvlm: Visual expert for pretrained language models. arXiv preprint arXiv:2311.03079 (2023)

    2023 arXiv

  23. [31]

    arXiv preprint arXiv:2312.09245 (2023)

    Wang, W., et al.: Drivemlm: Aligning multi-modal large language models with be- havioral planning states for autonomous driving. arXiv preprint arXiv:2312.09245 (2023)

    2023

  24. [32]

    arXiv preprint arXiv:2201.11903 (2022)

    Wei, J., Wang, X., Schuurmans, D., Bosma, M., Chi, E.H., Le, Q.V., et al.: Chain of thought prompting elicits reasoning in large language models. arXiv preprint arXiv:2201.11903 (2022)

    2022 arXiv

  25. [33]

    arXiv preprint arXiv:2309.16292 (2023)

    Wen, L., et al.: Dilu: A knowledge-driven approach to autonomous driving with large language models. arXiv preprint arXiv:2309.16292 (2023)

    2023 arXiv

  26. [34]

    arXiv preprint arXiv:2312.11935 (2023)

    Xia, Y., et al.: Parameterized decision-making with multi-modal perception for autonomous driving. arXiv preprint arXiv:2312.11935 (2023)

    2023 arXiv

  27. [35]

    arXiv preprint arXiv:2310.01412 (2023)

    Xu, Z., et al.: Drivegpt4: Interpretable end-to-end autonomous driving via large language model. arXiv preprint arXiv:2310.01412 (2023)

    2023 arXiv

  28. [36]

    arXiv e-prints arXiv:2311 (2023)

    Yang, Z., et al.: Llm4drive: A survey of large language models for autonomous driving. arXiv e-prints arXiv:2311 (2023)

    2023

  29. [37]

    arXiv preprint arXiv:2306.13549 (2023)

    Yin, S., et al.: A survey on multimodal large language models. arXiv preprint arXiv:2306.13549 (2023)

    2023 arXiv

  30. [38]

    arXiv preprint arXiv:2401.13601 (2024)

    Zhang, D., et al.: Mm-llms: Recent advances in multimodal large language models. arXiv preprint arXiv:2401.13601 (2024)

    2024 arXiv

  31. [39]

    In: International Conference on Neural Information Processing

    Zhang, L., Li, R.: Knowledge prompting with contrastive learning for unsupervised commonsenseqa. In: International Conference on Neural Information Processing. pp. 27–38. Springer (2023)

    2023

  32. [40]

    arXiv preprint arXiv:2304.10592 (2023)

    Zhu, D., et al.: Minigpt-4: Enhancing vision-language understanding with advanced large language models. arXiv preprint arXiv:2304.10592 (2023)

    2023 arXiv

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.