Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

ALN-P3: Unified Language Alignment for Perception, Prediction, and Planning in Autonomous Driving

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read A training-only cross-modal alignment across perception, prediction, and planning cuts collisions and improves driving language reasoning.

desk verdict Incremental but plausible: training-only alignment losses yield strong single-run numbers; the planning-loss grounding claim needs work. read the letter →

arxiv 2505.15158 v1 pith:OJHF22VG submitted 2025-05-21 cs.CV cs.CL

Yunsheng Ma , Burhaneddin Yaman , Xin Ye , Mahmut Yurt , Jingru Luo , Abhirup Mallik , Ziran Wang , Liu Ren This is my paper · ORCID
classification cs.CVcs.CL
keywords autonomousdrivingend-to-endplanningvision-languagemodelscross-modalalignmentco-distillationBEVperceptionexplanationvisualquestionanswering
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper claims that an autonomous-driving system can drive better and explain itself better if its visual representations are explicitly aligned with language at every stage of the perception–prediction–planning stack, rather than only at the final output. Its framework, ALN-P3, is a training-only co-distillation strategy that adds three alignment losses—one for object perception, one for agent motion prediction, and one for ego planning—between a fast vision-based planner and a slow multimodal language model. Because the alignment modules are active only during training, the deployed system keeps the fast planner's inference cost. The reported results show the average collision rate on nuScenes falling from 0.22% to 0.16%, a 27% relative improvement, alongside better language reasoning on Nu-X, TOD3Cap, and nuScenes-QA than prior baselines. A sympathetic reading is that language supervision can regularize the whole driving stack with no runtime penalty, which is the payoff a real-world deployment would care about.

What carries the argument

The argument is carried by a co-distillation objective made of three alignment losses, one per stage of the P3 stack. P1A uses mean squared error between an MLP-projected instance feature and a frozen CLIP text embedding of the corresponding object caption. P2A and P3A use an attention-based pooling operator that maps heterogeneous inputs—continuous trajectory waypoints or discrete MLLM output logits—into a shared space spanned by learnable prompt tokens; P2A then applies a CLIP-style contrastive loss between pooled agent trajectories and pooled language logits, while P3A applies a negative cosine similarity loss between the pooled ego plan and the pooled planning-response logits. These modules sit between the fast planner's intermediate tokens and the slow MLLM's output distribution, so training updates both branches, and they are removed at inference, adding no cost to deployment.

What would settle it

Train ALN-P3 on nuScenes with P3A pairs shuffled so each ego trajectory is aligned with a planning-response logit from another scene; if the shuffled model keeps the same average collision rate and Nu-X CIDEr as the correctly paired model, the planning-alignment loss carries no semantic information and the central claim fails.

Watch

Extended reading notes

Core claim

The central claim is that cross-modal consistency between what the fast driving stack perceives and what the slow language model says is itself a useful training signal for both branches. ALN-P3 introduces three alignment modules: P1A aligns instance-level bird's-eye-view features with CLIP text embeddings of ground-truth object captions through an MSE loss; P2A projects predicted agent trajectories and the MLLM's output logits into a shared space of learnable prompt tokens via attention-based pooling and applies a CLIP-style contrastive loss; P3A projects the ego plan and planning-related language logits into the same kind of shared space and minimizes their negative cosine similarity. All three modules are applied only during training and are selected per example according to the DriveLM prompt category. On nuScenes, the method reports collision rates of 0.05%, 0.09%, and 0.35% at 1s, 2s, and 3s horizons, an average of 0.16%; on language benchmarks it reports CIDEr 28.6 on Nu-X, CIDEr 341.9 on TOD3Cap, and 52.9% accuracy on nuScenes-QA. The paper concludes that aligning perception, prediction, and planning with language simultaneously improves driving decisions and language reasoning.

Load-bearing premise

The load-bearing premise is that the shared prompt-token space is semantically coherent, so making a plan embedding and a language embedding closer together really means the plan matches the words; if that space is arbitrary, P3A is just an ungrounded regularizer and the safety gains would not demonstrate language-grounded planning.

Editorial extensions

If this is right

  • A single training-time alignment signal improves both driving safety and language reasoning, so the two objectives need not compete.
  • The fast planner's inference cost is unchanged, meaning the alignment recipe can be added to an existing end-to-end driving stack without a latency trade-off.
  • Grounding language generation in intermediate driving representations beats using general-purpose multimodal models on driving-specific explanation and QA benchmarks.
  • The same three losses cover perception, prediction, and planning queries via prompt-category selection, so one framework handles the full P3 stack.
  • State-of-the-art results on four benchmarks suggest the benefit transfers across planning, explanation, dense captioning, and visual question answering.

Reading between the lines

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

  • If the alignment losses are the true cause of the gains, the same co-distillation recipe could extend to closed-loop driving or world-model pretraining, where language descriptions could regularize latent traffic states with no deployment cost.
  • The P3A objective presumes the learned prompt-token space is semantically coherent; a natural stress test is to rank matched versus mismatched trajectory–language pairs by cosine similarity and to compare against a contrastive loss variant.
  • The method depends on category-specific prompt routing, an admitted limitation, so open-ended or compositional queries may not benefit equally; testing dynamic prompt selection would delimit where the gain comes from.
  • The collision-rate improvement is measured in open-loop planning on nuScenes; closed-loop evaluation would show whether alignment-induced behavioral differences persist under reactive traffic.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes ALN-P3, a training-only co-distillation framework that inserts three cross-modal alignment losses between a "fast" vision-based perception–prediction–planning (P3) stack and a "slow" multimodal LLM module. P1A aligns instance-level BEV features with CLIP text embeddings of ground-truth object captions via MSE; P2A aligns predicted agent trajectories with MLLM output logits via a CLIP-style contrastive loss over learned prompt tokens; P3A aligns the ego trajectory with planning-related MLLM logits via a negative cosine similarity loss. All alignment losses are disabled at inference. Experiments on nuScenes, Nu-X, TOD3Cap, and nuScenes-QA report state-of-the-art collision rates (0.05%, 0.09%, 0.35% at 1s/2s/3s) and strong language reasoning scores (CIDEr 28.6 on Nu-X, 341.9 on TOD3Cap, 52.9% on nuScenes-QA).

Significance. The training-only design is practically attractive because it adds no inference cost, and a complete perception–prediction–planning language alignment is a useful goal. If the reported numbers are reproducible, the improvements over Hint-AD on Nu-X and TOD3Cap are substantial, and the paper would demonstrate that cross-modal alignment can benefit both planning safety and language reasoning simultaneously. However, the current evidence is a single run without error bars or code, there are no ablations isolating the three alignment modules, and the P3A loss in Eq. (8) is not the scale-invariant negative cosine similarity the text claims. The central claim that P3A semantically grounds planning in language is therefore not yet established, although the issues appear addressable with additional experiments and a corrected formulation.

major comments (4)
  1. [§3.4, Eq. (8)] The P3A loss is not scale-invariant negative cosine similarity: the denominator is ||z_plan||_2^2 · ||z_llm||_2^2, whereas negative cosine similarity requires ||z_plan||_2 · ||z_llm||_2. For positively correlated vectors, minimizing this loss favors shrinking the norms, inviting scale collapse, so the objective actually optimized is not the semantic similarity described in the text. This is load-bearing for the planning-alignment claim, and the equation must be corrected and the experiments re-run or re-analyzed with the intended loss.
  2. [§3.4, Eq. (7)] Unlike P1A, which anchors to frozen CLIP text embeddings, and P2A, which uses contrastive pairs, P3A has no fixed semantic target: both z_plan and z_llm are obtained by attention pooling over the same learnable prompt tokens P3, and the language side is MLLM output logits rather than a text embedding. The loss can be minimized by making the two pooled representations agree without encoding any planning-language semantics, so the claim that P3A grounds ego planning in language is not supported. The paper should provide an ablation isolating P3A and an analysis of the learned prompt space, for example by inspecting nearest text embeddings or comparing against a variant with random prompt tokens.
  3. [§4.4, Table 1] The experimental evidence consists of a single run with no error bars, no code, and no ablations of the three alignment modules. The collision-rate differences between the proposed method and baselines (e.g., 0.05% vs. 0.09% at 1s) may be within run-to-run variance, and the improvement over Hint-AD is not quantified statistically. Likewise, without ablations that remove P1A, P2A, and P3A individually, the reported gains cannot be attributed to the proposed alignment mechanisms. At minimum, the authors should report multiple seeds with standard deviations or confidence intervals and add per-module ablations.
  4. [§4.3] The statement that all loss terms are equally weighted with a default weight of 1 is not a neutral choice: MSE, contrastive, and cosine losses have very different scales, and the reported results may be sensitive to this arbitrary weighting. Since the paper explicitly avoids tuning due to resource constraints, a small sensitivity analysis over the alignment-loss weights is needed to establish that the central results are not an artifact of the chosen scales.
minor comments (5)
  1. [§3.2, Eq. (3)] P1A uses MSE on CLIP embeddings, but the text does not state whether the projected visual features and text embeddings are normalized before computing the loss; with CLIP embeddings the scale matters, so please clarify.
  2. [§3.3, Eqs. (4)–(6)] The attention-pooling operator in Eq. (4) is defined with a generic sum over N, but its application in Eq. (5) uses trajectories V_a of shape T_f×2 and MLLM output logits O whose exact arrangement for the Na agents is not specified; please clarify how the pooling indices align with agents and prompt tokens.
  3. [Table 1] The row label "P3 (Ours)" is inconsistent with the paper's naming and should be "ALN-P3"; the formatting of the collision-rate row for DriveVLM and the spacing in "V AD" should also be fixed.
  4. [§2.2] There is a double period after "(Tian et al., 2024b).." and the reference list should be checked for completeness (for example, the DriveLM taxonomy used in §3.4 is cited but not listed).
  5. [Tables 2–4] The qualitative examples are individual samples and do not by themselves demonstrate that the decoded outputs "align closely" with ground truth; a small quantitative error analysis or human evaluation would make the claim more credible.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: alignment losses are training regularizers independent of the evaluation protocol; the single self-citation is not load-bearing.

full rationale

The paper's derivation chain is self-contained with respect to circularity. P1A (Eq. 3) projects instance-level BEV features and minimizes MSE against CLIP text embeddings of ground-truth captions; P2A (Eqs. 5-6) and P3A (Eqs. 7-8) define contrastive and cosine-style losses between projected trajectory/logit features. None of these alignment losses is a function of the reported metrics (collision rate, CIDEr, BLEU, METEOR, ROUGE-L, nuScenes-QA accuracy); the evaluated quantities come from planner outputs and autoregressive language decoding, not from the alignment terms themselves. The self-citation to Ma et al. 2025 appears only as a related-work pointer in the introduction and carries no load in the method or evaluation. The technical concerns raised by the reader -- Eq. 8 has squared norms in the denominator, and the P3A shared space is learned without a fixed text anchor -- are correctness and robustness objections to whether the loss achieves semantic grounding, not circular reductions; they do not make the prediction equal to the input. The absence of an ablation isolating P3A is an evidence gap, not a circularity. Overall circularity score: 0.

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

The method introduces no new physical or architectural entities beyond learned prompt tokens and projection heads, which are standard components. The main unverified assumptions concern the semantic validity of the shared embedding spaces used by the alignment losses.

free parameters (3)
  • loss_weights = 1.0 for all losses
    Section 4.3 states all losses are equally weighted with default weight 1 and no tuning; this is a human-chosen hyperparameter that could affect the balance.
  • prompt_token_counts_N2_N3 = not reported
    The sizes of learnable prompt token sets in P2A and P3A (Sec 3.3, 3.4) are not given; they determine the capacity of the shared alignment space.
  • projection_heads_Phi = trained
    The MLPs that map visual and language features into the shared space are learned and are central to the alignment; their capacity is not analyzed.
assumptions (3)
  • domain assumption BEV instance features and CLIP text embeddings are comparable via MSE after a linear projection
    P1A (Eq. 3) assumes that Euclidean distance in the projected space reflects semantic similarity.
  • domain assumption Contrastive loss between trajectory and language logits is beneficial
    The P2A (Eq. 6) assumes that pulling trajectory embeddings and language embeddings together in a shared space improves both tasks.
  • domain assumption The MLLM output logits can be pooled to a single vector via attention-based pooling in a meaningful way
    P2A and P3A use the attention pooling A (Eq. 4) to compress a sequence of logits into a single embedding without validation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ALN-P3: Unified Language Alignment for Perception, Prediction, and Planning in Autonomous Driving." pith.science (2026). https://pith.science/paper/OJHF22VG

@misc{pith2026250515158,
  author       = {Pith},
  title        = {Pith review of: ALN-P3: Unified Language Alignment for Perception, Prediction, and Planning in Autonomous Driving},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OJHF22VG}},
  note         = {Machine review of arXiv:2505.15158}
}
read the original abstract

Recent advances have explored integrating large language models (LLMs) into end-to-end autonomous driving systems to enhance generalization and interpretability. However, most existing approaches are limited to either driving performance or vision-language reasoning, making it difficult to achieve both simultaneously. In this paper, we propose ALN-P3, a unified co-distillation framework that introduces cross-modal alignment between "fast" vision-based autonomous driving systems and "slow" language-driven reasoning modules. ALN-P3 incorporates three novel alignment mechanisms: Perception Alignment (P1A), Prediction Alignment (P2A), and Planning Alignment (P3A), which explicitly align visual tokens with corresponding linguistic outputs across the full perception, prediction, and planning stack. All alignment modules are applied only during training and incur no additional costs during inference. Extensive experiments on four challenging benchmarks-nuScenes, Nu-X, TOD3Cap, and nuScenes QA-demonstrate that ALN-P3 significantly improves both driving decisions and language reasoning, achieving state-of-the-art results.

Figures

Figures reproduced from arXiv: 2505.15158 by the authors.

Figure 1
Figure 1. Overview of the proposed ALN-P3 framework, which integrates vision and language alignment across the full autonomous driving stack. The architecture includes three alignment modules: Perception Alignment (P1A), Prediction Alignment (P2A), and Planning Alignment (P3A). These modules align BEV-based visual tokens, such as instance, motion, and ego features, with corresponding natural language outputs through cross-mod… view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. A User-driven Design Framework for Robotaxi

    cs.RO 2026-02 conditional novelty 5.0 of 10

    Real-world Chinese robotaxi users value driverless privacy and consistent driving but struggle with rigid stops, poor transparency, and unclear emergency controls; these findings support a four-phase, user-driven desi...

Reference graph

Works this paper leans on

35 extracted references · 11 canonical work pages · cited by 1 Pith paper

  1. [1]

    Satanjeev Banerjee and Alon Lavie. 2005. METEOR : An Automatic Metric for MT Evaluation with Improved Correlation with Human Judgments . In ACL Workshop on Intrinsic and Extrinsic Evaluation Measures for Machine Translation and/or Summarization

    work page 2005

  2. [2]

    Lang, Sourabh Vora, Venice Erin Liong, Qiang Xu, Anush Krishnan, Yu Pan, Giancarlo Baldan, and Oscar Beijbom

    Holger Caesar, Varun Bankiti, Alex H. Lang, Sourabh Vora, Venice Erin Liong, Qiang Xu, Anush Krishnan, Yu Pan, Giancarlo Baldan, and Oscar Beijbom. 2020. nuScenes : A Multimodal Dataset for Autonomous Driving . In CVPR

    work page 2020

  3. [3]

    Li Chen, Penghao Wu, Kashyap Chitta, Bernhard Jaeger, Andreas Geiger, and Hongyang Li. 2024 a . https://doi.org/10.1109/TPAMI.2024.3435937 End-to-end Autonomous Driving : Challenges and Frontiers . TPAMI

    arXiv 2024

  4. [4]

    Zhe Chen, Jiannan Wu, Wenhai Wang, Weijie Su, Guo Chen, Sen Xing, Muyan Zhong, Qinglong Zhang, Xizhou Zhu, Lewei Lu, Bin Li, Ping Luo, Tong Lu, Yu Qiao, and Jifeng Dai. 2024 b . https://doi.org/10.48550/arXiv.2312.14238 InternVL : Scaling up Vision Foundation Models and Aligning for Generic Visual - Linguistic Tasks . In CVPR

    arXiv doi:10.48550/arxiv.2312.14238 2024

  5. [5]

    Kairui Ding, Boyuan Chen, Yuchen Su, Huan-ang Gao, Bu Jin, Chonghao Sima, Wuqiang Zhang, Xiaohui Li, Paul Barsch, Hongyang Li, and Hao Zhao. 2024. Hint- AD : Holistically Aligned Interpretability in End -to- End Autonomous Driving . In CoRL

    work page 2024

  6. [6]

    Google . 2024. https://doi.org/10.48550/arXiv.2312.11805 Gemini: A Family of Highly Capable Multimodal Models . arXiv

    arXiv doi:10.48550/arxiv.2312.11805 2024

  7. [7]

    Patel, and Fatih Porikli

    Deepti Hegde, Rajeev Yasarla, Hong Cai, Shizhong Han, Apratim Bhattacharyya, Shweta Mahajan, Litian Liu, Risheek Garrepalli, Vishal M. Patel, and Fatih Porikli. 2025. https://doi.org/10.48550/arXiv.2501.09757 Distilling Multi -modal Large Language Models for Autonomous Driving . In CVPR

    arXiv doi:10.48550/arxiv.2501.09757 2025

  8. [8]

    Shengchao Hu, Li Chen, Penghao Wu, Hongyang Li, Junchi Yan, and Dacheng Tao. 2022. https://doi.org/10.48550/arXiv.2207.07601 ST - P3 : End -to-end Vision -based Autonomous Driving via Spatial - Temporal Feature Learning . In ECCV

    work page Pith review arXiv doi:10.48550/arxiv.2207.07601 2022

Show all 35 references
  1. [9]

    Yihan Hu, Jiazhi Yang, Li Chen, Keyu Li, Chonghao Sima, Xizhou Zhu, Siqi Chai, Senyao Du, Tianwei Lin, Wenhai Wang, Lewei Lu, Xiaosong Jia, Qiang Liu, Jifeng Dai, Yu Qiao, and Hongyang Li. 2023. https://doi.org/10.48550/arXiv.2212.10156 Planning-oriented Autonomous Driving . In CVPR

    2023 doi arXiv

  2. [10]

    Jyh-Jing Hwang, Runsheng Xu, Hubert Lin, Wei-Chih Hung, Jingwei Ji, Kristy Choi, Di Huang, Tong He, Paul Covington, Benjamin Sapp, James Guo, Dragomir Anguelov, and Mingxing Tan. 2024. https://doi.org/10.48550/arXiv.2410.23262 EMMA : End -to- End Multimodal Model for Autonomou...

    2024 doi arXiv

  3. [11]

    Xiaosong Jia, Junqi You, Zhiyuan Zhang, and Junchi Yan. 2025. https://doi.org/10.48550/arXiv.2503.07656 DriveTransformer : Unified Transformer for Scalable End -to- End Autonomous Driving . In ICLR

    2025 doi arXiv

  4. [12]

    Bo Jiang, Shaoyu Chen, Qing Xu, Bencheng Liao, Jiajie Chen, Helong Zhou, Qian Zhang, Wenyu Liu, Chang Huang, and Xinggang Wang. 2023. VAD : Vectorized Scene Representation for Efficient Autonomous Driving . In ICCV

    2023

  5. [13]

    Bu Jin, Yupeng Zheng, Pengfei Li, Weize Li, Yuhang Zheng, Sujie Hu, Xinyu Liu, Jinwei Zhu, Zhijie Yan, Haiyang Sun, Kun Zhan, Peng Jia, Xiaoxiao Long, Yilun Chen, and Hao Zhao. 2024. https://doi.org/10.1007/978-3-031-72649-1_21 TOD3Cap : Towards 3D Dense Captioning in Outdoor ...

    2024 doi

  6. [14]

    Muhammad Uzair Khattak, Hanoona Rasheed, Muhammad Maaz, Salman Khan, and Fahad Shahbaz Khan. 2023. https://doi.org/10.48550/arXiv.2210.03117 MaPLe : Multi -modal Prompt Learning . In CVPR

    2023 doi arXiv

  7. [15]

    Junnan Li, Dongxu Li, Silvio Savarese, and Steven Hoi. 2023. https://doi.org/10.48550/arXiv.2301.12597 BLIP -2: Bootstrapping Language - Image Pre -training with Frozen Image Encoders and Large Language Models . In ICML

    2023 doi arXiv

  8. [16]

    Zhiqi Li, Wenhai Wang, Hongyang Li, Enze Xie, Chonghao Sima, Tong Lu, Yu Qiao, and Jifeng Dai. 2022. https://doi.org/10.1007/978-3-031-20077-9_1 BEVFormer : Learning Bird ’s- Eye - View Representation from Multi - Camera Images via Spatiotemporal Transformers . In ECCV

    2022 doi

  9. [17]

    Zhiqi Li, Zhiding Yu, Shiyi Lan, Jiahan Li, Jan Kautz, Tong Lu, and Jose M. Alvarez. 2024. https://doi.org/10.48550/arXiv.2312.03031 Is Ego Status All You Need for Open - Loop End -to- End Autonomous Driving ? In CVPR

    2024 doi arXiv

  10. [18]

    Chin-Yew Lin. 2004. ROUGE : A Package for Automatic Evaluation of Summaries . In Text Summarization Branches Out

    2004

  11. [19]

    Yunsheng Ma, Wenqian Ye, Can Cui, Haiming Zhang, Shuo Xing, Fucai Ke, Jinhong Wang, Chenglin Miao, Jintai Chen, Hamid Rezatofighi, Zhen Li, Guangtao Zheng, Chao Zheng, Tianjiao He, Manmohan Chandraker, Burhaneddin Yaman, Xin Ye, Hang Zhao, and Xu Cao. 2025. Position: Prospecti...

    2025

  12. [20]

    OpenAI. 2023. https://doi.org/10.48550/arXiv.2303.08774 GPT -4 Technical Report . arXiv

    2023 doi arXiv

  13. [21]

    Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu. 2002. https://doi.org/10.3115/1073083.1073135 Bleu: a Method for Automatic Evaluation of Machine Translation . In ACL

    2002

  14. [22]

    Tianwen Qian, Jingjing Chen, Linhai Zhuo, Yang Jiao, and Yu-Gang Jiang. 2024. https://doi.org/10.1609/aaai.v38i5.28253 NuScenes - QA : A Multi - Modal Visual Question Answering Benchmark for Autonomous Driving Scenario . In AAAI

    2024 doi

  15. [23]

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. 2021. Learning Transferable Visual Models From Natural Language Supervision . In ICML

    2021

  16. [24]

    Katrin Renz, Long Chen, Elahe Arani, and Oleg Sinavski. 2025. https://doi.org/10.48550/arXiv.2503.09594 SimLingo : Vision - Only Closed - Loop Autonomous Driving with Language - Action Alignment . In CVPR

    2025 doi arXiv

  17. [25]

    Chonghao Sima, Katrin Renz, Kashyap Chitta, Li Chen, Hanxue Zhang, Chengen Xie, Ping Luo, Andreas Geiger, and Hongyang Li. 2024. https://doi.org/10.48550/arXiv.2312.14150 DriveLM : Driving with Graph Visual Question Answering . In ECCV

    2024 doi arXiv

  18. [26]

    Ran Tian, Boyi Li, Xinshuo Weng, Yuxiao Chen, Edward Schmerling, Yue Wang, Boris Ivanovic, and Marco Pavone. 2024 a . Tokenize the World into Object -level Knowledge to Address Long -tail Events in Autonomous Driving . In CoRL

    2024

  19. [27]

    Xiaoyu Tian, Junru Gu, Bailin Li, Yicheng Liu, Chenxu Hu, Yang Wang, Kun Zhan, Peng Jia, Xianpeng Lang, and Hang Zhao. 2024 b . https://doi.org/10.48550/arXiv.2402.12289 DriveVLM : The Convergence of Autonomous Driving and Large Vision - Language Models . In CoRL

    2024 doi arXiv

  20. [28]

    Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, Dan Bikel, Lukas Blecher, Cristian Canton Ferrer, Moya Chen, Guillem Cucurull, David Esiobu, Jude Fernandes, Jeremy Fu, W...

    2023 doi arXiv

  21. [29]

    Lawrence Zitnick, and Devi Parikh

    Ramakrishna Vedantam, C. Lawrence Zitnick, and Devi Parikh. 2015. https://doi.org/10.1109/CVPR.2015.7299087 CIDEr : Consensus -based image description evaluation . In CVPR

    2015

  22. [30]

    Shihao Wang, Zhiding Yu, Xiaohui Jiang, Shiyi Lan, Min Shi, Nadine Chang, Jan Kautz, Ying Li, and Jose M. Alvarez. 2025. https://doi.org/10.48550/arXiv.2405.01533 OmniDrive : A Holistic Vision - Language Dataset for Autonomous Driving with Counterfactual Reasoning . In CVPR

    2025 doi arXiv

  23. [31]

    Xinshuo Weng, Boris Ivanovic, Yan Wang, Yue Wang, and Marco Pavone. 2024. https://doi.org/10.1109/CVPR52733.2024.01463 PARA - Drive : Parallelized Architecture for Real -time Autonomous Driving . In CVPR

    2024

  24. [32]

    Sihyun Yu, Sangkyung Kwak, Huiwon Jang, Jongheon Jeong, Jonathan Huang, Jinwoo Shin, and Saining Xie. 2025. https://doi.org/10.48550/arXiv.2410.06940 Representation Alignment for Generation : Training Diffusion Transformers Is Easier Than You Think . In ICLR . Oral

    2025 doi arXiv

  25. [33]

    Renrui Zhang, Jiaming Han, Chris Liu, Peng Gao, Aojun Zhou, Xiangfei Hu, Shilin Yan, Pan Lu, Hongsheng Li, and Yu Qiao. 2024. LLaMA - Adapter : Efficient Fine -tuning of Language Models with Zero -init Attention . In ICLR

    2024

  26. [34]

    online" 'onlinestring :=

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint eprinttype howpublished institution journal key month note number organization pages publisher school series title type volume year doi pubmed url lastchecked label extra.label sort.label short.list...

  27. [35]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

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