Pith. sign in

REVIEW 3 major objections 4 minor 40 references

ObjVariantEnsemble: Advancing Point Cloud LLM Evaluation in Challenging Scenes with Subtly Distinguished Objects

T0 review · 3 major / 4 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 The paper's ObjVariantEnsemble benchmark assembles 3D scenes with deliberately similar distractor objects and finds that current point-cloud LLMs align textual descriptions to shape and class far better than to pure location.

desk verdict A useful benchmark framework with a load-bearing unverified assumption about annotation quality; worth reviewing, but the core spatial-reasoning claim needs stronger evidence. read the letter →

arxiv 2412.14837 v1 pith:TIDQTYDA submitted 2024-12-19 cs.CV

Qihang Cao , Huangxun Chen This is my paper · ORCID
classification cs.CV
keywords 3Dsceneunderstandingpointcloudgroundingbenchmarkconstructionspatialreasoningdistractorobjectsvision-languagemodelsLLMevaluation
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 argues that existing 3D scene benchmarks are too easy and too ambiguous to reveal what point-cloud LLMs actually understand, because their text annotations often fail to single out one object. To fix this, it builds ObjVariantEnsemble (OVE), a pipeline that places a target object next to intentionally similar distractors and generates annotations naming the specific visual or spatial difference that isolates the target. The paper claims that OVE exposes a real weakness: state-of-the-art 3D grounding models use shape, class, and color cues effectively, but their grounding performance collapses when the only distinguishing information is the target's location. If true, this gives a concrete target for improving spatial reasoning in 3D representations and position encodings.

What carries the argument

The central mechanism is the OVE assembly and annotation pipeline: an object retriever pulls distractors from object-level datasets that match the target in class, color, or shape according to a chosen similarity dimension, and a scene integrator places them at prescribed spatial offsets (left, right, front, back, up, down, and combinations forming 13 predicates) inside a real scanned scene background. A distinction recognizer then renders multi-view images and runs an iterative loop where an LLM asks and a VLM answers about potential differences, and a summarizer LLM compresses the answers into one fine-grained annotation. This machinery turns 'challenging scene' into a controllable experimental variable: the number of distractors sets difficulty, and the distinction type (location, location+shape, location+color, location+class) determines which cue the model must rely on.

What would settle it

Take a fixed set of OVE scenes and redo the location-only condition with human-written annotations or with multiple independently generated LLM/VLM annotations, then measure inter-annotation agreement and grounding accuracy; if accuracy jumps or agreement is low and uneven across attributes, the conclusion that models are intrinsically weak at pure location reasoning does not hold.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central discovery is that when a scene contains several similar objects, the cue that lets a model find the right one is mostly visual identity (shape, class, color), not geometry: grounding accuracy on OVE drops sharply when the only distinguishing cue is location, and a model with explicit location encoding does not beat the baseline in those cases. The paper attributes this to the way 3D encoders are pretrained on object-level point clouds, which teaches shape and category features but not relational spatial features. It reinforces the point with a segmentation study showing that distractors degrade even purely visual encoders, and that same-class distractors are less harmful than different-class ones, suggesting the weakness lies partly in the 3D features themselves rather than only in text alignment.

Load-bearing premise

The entire benchmark rests on the assumption that the LLM/VLM-generated annotations are accurate, consistent, and equally reliable across attributes, so that the relative difficulty of location versus shape and color reflects the models' abilities rather than the annotations' quality.

Editorial extensions

If this is right

  • Grounding accuracy on OVE declines as the number of distractors increases, so the benchmark can rank models by robustness rather than by performance on single-scan scenes.
  • If location is the weakest cue, position embeddings and spatial token designs in 3D LLMs need rethinking rather than simple extension with more coordinate data.
  • Fine-tuning a model on OVE scenes improves grounding, indicating that distractor-rich training data is a direct lever for better 3D understanding.
  • The segmentation results show that distractors hurt even pure visual encoders, so the spatial deficit is not only a text-alignment problem but also a feature-learning problem.
  • The same pipeline can be extended to counting, captioning, and question answering, allowing the construction scheme to generate future benchmarks beyond grounding.

Reading between the lines

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

  • The headline result is conditional on annotation quality: the paper reports continuous manual verification but gives no agreement rates or error counts, so if location annotations are systematically noisier or less informative than shape and color ones, the spatial-reasoning deficit could be partly a benchmark artifact.
  • A natural next probe is to vary how the same spatial relation is phrased (for example, 'left of' versus 'west of' versus 'near') to separate genuine geometric reasoning from language-alignment failures.
  • Because OVE mostly uses point-only objects without mesh data, models that rely on 2D renders or texture are disadvantaged; generating mesh data from points would make the benchmark fairer for such models.
  • The assembly procedure could be repurposed as a data augmentation engine for training, generating unlimited hard negative scenes instead of only evaluation scenes.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The paper introduces ObjVariantEnsemble (OVE), a benchmark construction scheme for point cloud scene-level grounding. It assembles scenes by inserting a target object and several distractors sampled from object-level datasets (ModelNet, ShapeNet, ScanObjectNN, OmniObject3D) into real ScanNet backgrounds, controlling similarity in class, color, shape, and spatial arrangement. An LLM-VLM cooperative annotator renders multi-view images, runs iterative QA rounds, and summarizes distinguishing attributes into natural-language annotations, with spatial relations added from integration specifications. The resulting benchmark contains roughly 75k scene-text pairs across four distinction types (location, location+shape, location+color, location+class). The authors evaluate Chat-3D-v2, a LoRA-fine-tuned variant, and M3dRefCLIP on 3D grounding, and also test Uni3D and PointBERT on object part segmentation with resampled scenes. They report that grounding accuracy decreases with more distractors and, in particular, that the models' performance with location-only distinctions is far below their performance with shape, color, or class distinctions.

Significance. OVE addresses a real gap: existing ScanNet-based benchmarks have fixed scenes and limited numbers of subtly confusable objects, and OVE's controlled ensembling allows challenge-level customization. The automated annotation pipeline is a useful step, and the decomposition by distinction type is a sensible diagnostic that can reveal whether models actually use spatial cues. The paper also honestly notes its current limits (e.g., no mesh data for texture-rich 2D rendering) in the conclusion. However, the headline empirical claim about location reasoning is not yet supported. The manuscript does not provide quantitative annotation-quality evidence, an explicit train/eval split for the fine-tuned model, or error bars on the main comparisons. These gaps are fixable and should be addressed before the evaluation conclusions are relied upon.

major comments (3)
  1. [§3.3, §4.2] The central conclusion of §4.2—that models are far worse at grounding by location than by shape, color, or class—assumes that location annotations are exactly as informative and uniquely identifying as visual-feature annotations. Location annotations are synthesized from integration specifications, while visual distinctions come from multi-round VLM Q&A, and the paper reports only 'continuously sampled for manual verification' with no sample size, inter-annotator agreement, per-type uniqueness rate, or error analysis. If location phrases such as 'surrounded by four chairs' or 'right side of the room, close to the door' are ambiguous or noisy in the assembled scenes, the Figure 6b gap could be an annotation artifact rather than evidence about spatial reasoning. Please provide per-type annotation-quality statistics and recompute the comparison on a subset verified to be uniquely identifying.
  2. [§4.1, §4.2] The evaluation of the fine-tuned Chat-3D-v2 uses 'a portion of our dataset for fine-tuning,' but the manuscript does not specify the train/eval split, the number of scenes used, or any leakage check. If the evaluation scenes overlap with the fine-tuning scenes, the reported improvements and the cross-type location-vs-visual comparison involving this model are not interpretable. Please state the exact split and confirm that the evaluation set is disjoint from training.
  3. [Figures 6 and 7] The main quantitative claims—decreasing accuracy with more distractors and the large gap between location-only and visual-feature conditions—are presented without error bars, confidence intervals, or the number of scenes per condition. With unknown variance, the 'far inferior' claim is not statistically grounded. Please report scene counts and variance estimates for each curve and condition.
minor comments (4)
  1. [References] ModelNet is attributed to (Sun et al. 2022), but that reference is a robustness-evaluation paper, not the ModelNet dataset; please cite the original ModelNet publication (Wu et al. 2015).
  2. [§3.3, Algorithm 1] The prompts named SUM P2 and SUM P3 are referenced but not included; please provide them in an appendix or in the project page for reproducibility.
  3. [Figures] Figures 6 and 7 would benefit from larger axis labels and a clear statement of what each plotted point represents; Figure 5's scene counts should also be tabulated unambiguously.
  4. [§5] The conclusion's limitation about lacking mesh data and rich texture is useful, but it should be stated earlier in the evaluation section where 2D-rendering-based models are excluded.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the benchmark evaluation is empirically self-contained and not forced by construction.

full rationale

No circular step is present. OVE constructs scenes by placing target and distractor objects according to integration specifications and generates annotations from those specifications plus LLM/VLM output (Sections 3.2-3.3). The evaluation in Section 4 feeds point cloud scenes and text annotations to external models and measures bounding-box IoU; nothing is fitted to the test data and no parameter is renamed as a prediction. The central Section 4.2 conclusion, that models align location text worse than shape, color, or class text, is an empirical observation about the tested models on a benchmark; it is not equivalent to the annotation procedure by construction, because model outputs are not derived from the integration specifications. The segmentation study in Section 4.3 fine-tunes models on OVE-style data and evaluates on similar data, but that is a standard experimental protocol rather than circularity. The validity concern that location annotations are synthesized from integration specifications while visual annotations come from VLM QA (Section 3.3: 'we further enhance the spatial location information based on the integration specifications' and 'We continuously sample annotations for manual verification to ensure quality') is a real annotation-quality threat to the strength of the Section 4.2 claim, but it is a correctness and validity issue, not a circularity issue: even if location annotations were noisier or less informative, the paper's argument would be empirically questionable, not true by definition. No self-citation chain, imported uniqueness theorem, or ansatz smuggled via citation appears. The benchmark's self-contained evaluation against external models supports a low circularity score; I assign 0.

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

The benchmark construction rests on the reliability of external 3D datasets and generative LLM/VLM models, plus several hand-chosen hyperparameters that shape the dataset and the evaluation results. The central claims are empirical rather than derived, so the ledger captures assumptions and manual settings rather than fitted physical parameters.

free parameters (4)
  • QA rounds = 6-7
    Number of iterative LLM-VLM Q&A rounds in Algorithm 1, chosen by hand to balance annotation richness and cost; directly affects annotation quality.
  • Distractor count range = 2 to 10
    The number of distractors used to vary challenge level in evaluation; chosen as an experimental variable, not fitted, but affects reported performance trends.
  • Shape similarity vocabulary = cuboids, L-shapes, spheres
    Only standard shapes are used to define shape similarity in the Object Retriever, a manual modeling choice that restricts the space of possible distractors.
  • Spatial primitive set = 6 primitives, 13 predicates
    The set of spatial relations (left, right, front, back, up, down, and combinations) is a design choice that determines the location annotations and the challenge distribution.
assumptions (4)
  • domain assumption Object-level datasets (ModelNet, ShapeNet, ScanObjectNN, OmniObject) provide accurate class, shape, and color labels
    The Object Retriever relies on these labels to select distractors; incorrect labels would propagate to distractor selection and downstream annotations.
  • domain assumption LLM (GPT) and VLM (LLaVA) outputs are reliable enough for annotation generation
    Section 3.3 depends on LLaVA and GPT to identify visual distinctions and summarize text; the paper provides only sampled manual verification, no quantitative accuracy metrics.
  • domain assumption ScanNet background scenes and target segmentations are accurate and suitable for ensembled objects
    The Scene Integrator uses original ScanNet backgrounds and target segmentation; errors would create unrealistic scenes and affect bounding-box-based evaluation.
  • domain assumption Manual screening of assembled scenes is sufficient to ensure visual coherence
    The paper states that assembled scenes were manually screened without identifying odd aspects, but no screening protocol or inter-rater agreement is reported.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ObjVariantEnsemble: Advancing Point Cloud LLM Evaluation in Challenging Scenes with Subtly Distinguished Objects." pith.science (2026). https://pith.science/paper/TIDQTYDA

@misc{pith2026241214837,
  author       = {Pith},
  title        = {Pith review of: ObjVariantEnsemble: Advancing Point Cloud LLM Evaluation in Challenging Scenes with Subtly Distinguished Objects},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TIDQTYDA}},
  note         = {Machine review of arXiv:2412.14837}
}
read the original abstract

3D scene understanding is an important task, and there has been a recent surge of research interest in aligning 3D representations of point clouds with text to empower embodied AI. However, due to the lack of comprehensive 3D benchmarks, the capabilities of 3D models in real-world scenes, particularly those that are challenging with subtly distinguished objects, remain insufficiently investigated. To facilitate a more thorough evaluation of 3D models' capabilities, we propose a scheme, ObjVariantEnsemble, to systematically introduce more scenes with specified object classes, colors, shapes, quantities, and spatial relationships to meet model evaluation needs. More importantly, we intentionally construct scenes with similar objects to a certain degree and design an LLM-VLM-cooperated annotator to capture key distinctions as annotations. The resultant benchmark can better challenge 3D models, reveal their shortcomings in understanding, and potentially aid in the further development of 3D models.

Figures

Figures reproduced from arXiv: 2412.14837 by the authors.

Figure 1
Figure 1. Comparison of 3D grounding benchmarks in chal [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. OVE Benchmark Construction Overview. distractors, summarizing their distinctions in terms of multi￾ple aspects (color/shape/location/...) as the annotation, e.g., the text in [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. ObjVariantEnsemble Scene Data Generation Framework. (For clearer illustration here, we only plot the target and [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Process for Capturing Annotations with Key Distinguishing Information. We render multi-view images and use [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: OVE Benchmark Summary port four different distinction types between the target object and distractors: location, location+shape, location+color, and location+class. Compared to existing scene-level 3D datasets, our work substantially expands the semantic richness of 3D…
Figure 4
Figure 4. Figure 4: It is worth mentioning that during the distinction [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 6
Figure 6. Figure 6: Experiments conducted on the OVE benchmark. [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]
Figure 7
Figure 7. Figure 7: Performance(mIoUI(%)) of 3D representation learning models on segmentation tasks using data resampled by OVE framework. 3D grounding, while Multi3DRef takes the initiative in ad￾dressing multi-object grounding. Both models utilize seg￾mentation as an auxiliary task to …
Figure 8
Figure 8. Figure 8: Visualizations of Segmentation Results on ShapeNet(Chang et al. 2015) Resampled: row (a) represents Uni3d, row [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

40 extracted references · 14 canonical work pages

  1. [1]

    Achlioptas, P.; Abdelreheem, A.; Xia, F.; Elhoseiny, M.; and Guibas, L. 2020. Referit3d: Neural listeners for fine-grained 3d object identification in real-world scenes. In Computer Vision--ECCV 2020: 16th European Conference, Glasgow, UK, August 23--28, 2020, Proceedings, Part I 16, 422--440. Springer

    2020

  2. [2]

    Azuma, D.; Miyanishi, T.; Kurita, S.; and Kawanabe, M. 2022. Scanqa: 3d question answering for spatial scene understanding. In proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 19129--19139

    work page 2022

  3. [3]

    Q.-H.; and Zhao, B

    Bai, F.; Du, Y.; Huang, T.; Meng, M. Q.-H.; and Zhao, B. 2024. M3d: Advancing 3d medical image analysis with multi-modal large language models. arXiv preprint arXiv:2404.00578

    arXiv 2024

  4. [4]

    X.; Funkhouser, T.; Guibas, L.; Hanrahan, P.; Huang, Q.; Li, Z.; Savarese, S.; Savva, M.; Song, S.; Su, H.; et al

    Chang, A. X.; Funkhouser, T.; Guibas, L.; Hanrahan, P.; Huang, Q.; Li, Z.; Savarese, S.; Savva, M.; Song, S.; Su, H.; et al. 2015. Shapenet: An information-rich 3d model repository. arXiv preprint arXiv:1512.03012

    arXiv 2015

  5. [5]

    Z.; Chang, A

    Chen, D. Z.; Chang, A. X.; and Nie ner, M. 2020. Scanrefer: 3d object localization in rgb-d scans using natural language. In European conference on computer vision, 202--221. Springer

    2020

  6. [6]

    Chen, S.; Chen, X.; Zhang, C.; Li, M.; Yu, G.; Fei, H.; Zhu, H.; Fan, J.; and Chen, T. 2024. LL3DA: Visual Interactive Instruction Tuning for Omni-3D Understanding Reasoning and Planning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 26428--26438

    work page 2024

  7. [7]

    X.; Savva, M.; Halber, M.; Funkhouser, T.; and Nie ner, M

    Dai, A.; Chang, A. X.; Savva, M.; Halber, M.; Funkhouser, T.; and Nie ner, M. 2017. Scannet: Richly-annotated 3d reconstructions of indoor scenes. In Proceedings of the IEEE conference on computer vision and pattern recognition, 5828--5839

    2017

  8. [8]

    Deitke, M.; Schwenk, D.; Salvador, J.; Weihs, L.; Michel, O.; VanderBilt, E.; Schmidt, L.; Ehsani, K.; Kembhavi, A.; and Farhadi, A. 2023. Objaverse: A universe of annotated 3d objects. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 13142--13153

    work page 2023

Show all 40 references
  1. [9]

    J.; Aydin, A.; Anvari, M.; Chakravarthy, A

    Ge, Y.; Tang, Y.; Xu, J.; Gokmen, C.; Li, C.; Ai, W.; Martinez, B. J.; Aydin, A.; Anvari, M.; Chakravarthy, A. K.; et al. 2024. BEHAVIOR Vision Suite: Customizable Dataset Generation via Simulation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recog...

    2024

  2. [10]

    Han, J.; Zhang, R.; Shao, W.; Gao, P.; Xu, P.; Xiao, H.; Zhang, K.; Liu, C.; Wen, S.; Guo, Z.; et al. 2023. Imagebind-llm: Multi-modality instruction tuning. arXiv preprint arXiv:2309.03905

    2023 arXiv

  3. [11]

    Hong, Y.; Zhen, H.; Chen, P.; Zheng, S.; Du, Y.; Chen, Z.; and Gan, C. 2023. 3d-llm: Injecting the 3d world into large language models. Advances in Neural Information Processing Systems, 36: 20482--20494

    2023

  4. [12]

    J.; Shen, Y.; Wallis, P.; Allen-Zhu, Z.; Li, Y.; Wang, S.; Wang, L.; and Chen, W

    Hu, E. J.; Shen, Y.; Wallis, P.; Allen-Zhu, Z.; Li, Y.; Wang, S.; Wang, L.; and Chen, W. 2021. Lora: Low-rank adaptation of large language models. arXiv preprint arXiv:2106.09685

    2021 arXiv

  5. [13]

    Huang, H.; Wang, Z.; Huang, R.; Liu, L.; Cheng, X.; Zhao, Y.; Jin, T.; and Zhao, Z. 2023 a . Chat-3D v2: Bridging 3D Scene and Large Language Models with Object Identifiers. arXiv preprint arXiv:2312.08168

    2023 arXiv

  6. [14]

    W.; Ouyang, W.; and Zuo, W

    Huang, T.; Dong, B.; Yang, Y.; Huang, X.; Lau, R. W.; Ouyang, W.; and Zuo, W. 2023 b . Clip2point: Transfer clip to point cloud classification with image-depth pre-training. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 22157--22167

    2023

  7. [15]

    Huang, W.; Wang, C.; Zhang, R.; Li, Y.; Wu, J.; and Fei-Fei, L. 2023 c . Voxposer: Composable 3d value maps for robotic manipulation with language models. arXiv preprint arXiv:2307.05973

    2023 arXiv

  8. [16]

    Jia, B.; Chen, Y.; Yu, H.; Wang, Y.; Niu, X.; Liu, T.; Li, Q.; and Huang, S. 2025. Sceneverse: Scaling 3d vision-language learning for grounded scene understanding. In European Conference on Computer Vision, 289--310. Springer

    2025

  9. [17]

    Jiang, L.; Zhao, H.; Shi, S.; Liu, S.; Fu, C.-W.; and Jia, J. 2020. Pointgroup: Dual-set point grouping for 3d instance segmentation. In Proceedings of the IEEE/CVF conference on computer vision and Pattern recognition, 4867--4876

    2020

  10. [18]

    Li, C.; Zhang, R.; Wong, J.; Gokmen, C.; Srivastava, S.; Mart \' n-Mart \' n, R.; Wang, C.; Levine, G.; Lingelbach, M.; Sun, J.; et al. 2023. Behavior-1k: A benchmark for embodied ai with 1,000 everyday activities and realistic simulation. In Conference on Robot Learning, 80--93. PMLR

    2023

  11. [19]

    Li, J.; Li, D.; Xiong, C.; and Hoi, S. 2022. Blip: Bootstrapping language-image pre-training for unified vision-language understanding and generation. In International conference on machine learning, 12888--12900. PMLR

    2022

  12. [20]

    Liu, H.; Li, C.; Wu, Q.; and Lee, Y. J. 2024 a . Visual instruction tuning. Advances in neural information processing systems, 36

    2024

  13. [21]

    Liu, M.; Shi, R.; Kuang, K.; Zhu, Y.; Li, X.; Han, S.; Cai, H.; Porikli, F.; and Su, H. 2024 b . Openshape: Scaling up 3d shape representation towards open-world understanding. Advances in Neural Information Processing Systems, 36

    2024

  14. [22]

    W.; Hallacy, C.; Ramesh, A.; Goh, G.; Agarwal, S.; Sastry, G.; Askell, A.; Mishkin, P.; Clark, J.; et al

    Radford, A.; Kim, J. W.; Hallacy, C.; Ramesh, A.; Goh, G.; Agarwal, S.; Sastry, G.; Askell, A.; Mishkin, P.; Clark, J.; et al. 2021. Learning transferable visual models from natural language supervision. In International conference on machine learning, 8748--8763. PMLR

    2021

  15. [23]

    Schult, J.; Engelmann, F.; Hermans, A.; Litany, O.; Tang, S.; and Leibe, B. 2023. Mask3d: Mask transformer for 3d semantic instance segmentation. In 2023 IEEE International Conference on Robotics and Automation (ICRA), 8216--8223. IEEE

    2023

  16. [24]

    Sun, J.; Zhang, Q.; Kailkhura, B.; Yu, Z.; Xiao, C.; and Mao, Z. M. 2022. Benchmarking robustness of 3d point cloud recognition against common corruptions. arXiv preprint arXiv:2201.12296

    2022 arXiv

  17. [25]

    Tang, Y.; Liu, J.; Wang, D.; Wang, Z.; Zhang, S.; Zhao, B.; and Li, X. 2024. Any2Point: Empowering Any-modality Large Models for Efficient 3D Understanding. arXiv preprint arXiv:2404.07989

    2024 arXiv

  18. [26]

    A.; Bauer, J.; Baumli, K.; Baveja, S.; Behbahani, F.; Bhoopchand, A.; Bradley-Schmieg, N.; Chang, M.; Clay, N.; Collister, A.; et al

    Team, A. A.; Bauer, J.; Baumli, K.; Baveja, S.; Behbahani, F.; Bhoopchand, A.; Bradley-Schmieg, N.; Chang, M.; Clay, N.; Collister, A.; et al. 2023. Human-timescale adaptation in an open-ended task space. arXiv preprint arXiv:2301.07608

    2023 arXiv

  19. [27]

    Touvron, H.; Martin, L.; Stone, K.; Albert, P.; Almahairi, A.; Babaei, Y.; Bashlykov, N.; Batra, S.; Bhargava, P.; Bhosale, S.; et al. 2023. Llama 2: Open foundation and fine-tuned chat models. arXiv preprint arXiv:2307.09288

    2023 arXiv

  20. [28]

    A.; Pham, Q.-H.; Hua, B.-S.; Nguyen, T.; and Yeung, S.-K

    Uy, M. A.; Pham, Q.-H.; Hua, B.-S.; Nguyen, T.; and Yeung, S.-K. 2019. Revisiting point cloud classification: A new benchmark dataset and classification model on real-world data. In Proceedings of the IEEE/CVF international conference on computer vision, 1588--1597

    2019

  21. [29]

    Wu, T.; Zhang, J.; Fu, X.; Wang, Y.; Ren, J.; Pan, L.; Wu, W.; Yang, L.; Wang, J.; Qian, C.; et al. 2023. Omniobject3d: Large-vocabulary 3d object dataset for realistic perception, reconstruction and generation. In Proceedings of the IEEE/CVF Conference on Computer Vision and ...

    2023

  22. [30]

    Xiao, T.; Chan, H.; Sermanet, P.; Wahid, A.; Brohan, A.; Hausman, K.; Levine, S.; and Tompson, J. 2022. Robotic skill acquisition via instruction augmentation with vision-language models. arXiv preprint arXiv:2211.11736

    2022 arXiv

  23. [31]

    Xu, R.; Wang, X.; Wang, T.; Chen, Y.; Pang, J.; and Lin, D. 2023. Pointllm: Empowering large language models to understand point clouds. arXiv preprint arXiv:2308.16911

    2023 arXiv

  24. [32]

    C.; et al

    Xue, L.; Yu, N.; Zhang, S.; Panagopoulou, A.; Li, J.; Mart \' n-Mart \' n, R.; Wu, J.; Xiong, C.; Xu, R.; Niebles, J. C.; et al. 2023. Ulip-2: Towards scalable multimodal pre-training for 3d understanding. arXiv preprint arXiv:2305.08275

    2023 arXiv

  25. [33]

    F.; and Chai, J

    Yang, J.; Chen, X.; Madaan, N.; Iyengar, M.; Qian, S.; Fouhey, D. F.; and Chai, J. 2024. 3D-GRAND: A Million-Scale Dataset for 3D-LLMs with Better Grounding and Less Hallucination. arXiv preprint arXiv:2406.05132

    2024 arXiv

  26. [34]

    Yu, X.; Tang, L.; Rao, Y.; Huang, T.; Zhou, J.; and Lu, J. 2022. Point-bert: Pre-training 3d point cloud transformers with masked point modeling. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 19313--19322

    2022

  27. [35]

    Zeng, Y.; Jiang, C.; Mao, J.; Han, J.; Ye, C.; Huang, Q.; Yeung, D.-Y.; Yang, Z.; Liang, X.; and Xu, H. 2023. Clip2: Contrastive language-image-point pretraining from real-world point cloud data. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognit...

    2023

  28. [36]

    Zhang, B.; Yuan, J.; Shi, B.; Chen, T.; Li, Y.; and Qiao, Y. 2023. Uni3d: A unified baseline for multi-dataset 3d object detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 9253--9262

    2023

  29. [37]

    Zhang, Y.; Gong, Z.; and Chang, A. X. 2023. Multi3drefer: Grounding text description to multiple 3d objects. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 15225--15236

    2023

  30. [38]

    Zheng, L.; Chiang, W.-L.; Sheng, Y.; Zhuang, S.; Wu, Z.; Zhuang, Y.; Lin, Z.; Li, Z.; Li, D.; Xing, E.; et al. 2024. Judging llm-as-a-judge with mt-bench and chatbot arena. Advances in Neural Information Processing Systems, 36

    2024

  31. [39]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all...

  32. [40]

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