Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

Dense360: Dense Understanding from Omnidirectional Panoramas

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 The paper claims that equirectangular panoramas, read through a position encoding that respects their wrap-around and latitude geometry, give multimodal language models dense 360° scene understanding.

desk verdict Dense360 builds a substantial panoramic dataset and benchmark, but the headline gains are undermined by a circular training/evaluation pipeline and a missing human-validated subset. read the letter →

arxiv 2506.14471 v1 pith:2TIXMAIH submitted 2025-06-17 cs.CV

Yikang Zhou , Tao Zhang , Dizhe Zhang , Shunping Ji , Xiangtai Li , Lu Qi This is my paper · ORCID
classification cs.CV
keywords omnidirectionalpanoramaequirectangularprojectiondensecaptioningreferringexpressiongroundingmultimodallargelanguagemodelpositionencodingreliability-scoredannotations360-degreebenchmark
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 multimodal language models can be taught dense understanding of 360° scenes by feeding them a single equirectangular panorama instead of many perspective views, provided the model's position encoding respects the geometry of that projection. To support the claim it contributes three things: a 160K-panorama dataset with 5M entity-level captions, 1M referring expressions, and 100K entity-grounded scene descriptions, each caption carrying an automatic reliability score; Dense360-Bench, a first benchmark for omnidirectional captioning and grounding; and ERP-RoPE, a position encoding that makes the horizontal axis wrap around and corrects for latitude-dependent pixel density. With ERP-RoPE and the dataset, a 3B model reaches 51.78 captioning and 76.81 grounding on the benchmark, above a 4B grounding-specialist model post-trained on the same data (47.80 and 74.39). If these numbers hold, a compact single-image input can replace multi-view feature fusion for panoramic scene understanding.

What carries the argument

ERP-RoPE is the load-bearing object: a rotary position encoding for equirectangular images that replaces the ordinary horizontal coordinate with a periodic, center-peaked coordinate f(w) and rescales it by gamma = H / sum(cos theta) so that all latitude circles receive a matched total positional budget. These two choices encode wrap-around continuity (the left and right borders are adjacent) and latitude-dependent information density (pixels near the poles are physically closer together). The paper plugs ERP-RoPE into the multimodal rotary position embedding of a vision-language model, adds a [SEG] token decoded into segmentation masks by SAM, and trains with LoRA on the Dense360 dataset, while perspective images and text keep the standard mRoPE encoding.

What would settle it

Re-annotate a random sample of Dense360-Bench entities with human verifiers, producing human-checked masks and captions, then recompute Dense360VLM's captioning and grounding scores on that subset. If agreement with the automatic labels is low, or if the model's margin over baselines shrinks on the human-validated subset, the benchmark would be measuring agreement with the generator rather than true dense understanding.

Watch

Extended reading notes

Core claim

On its own terms, the central discovery is that the two geometric facts of equirectangular projection—horizontal continuity, where the image's left and right edges are adjacent in the scene, and latitude-dependent distortion, where a pixel row near a pole covers less physical circumference than the equator—are not handled by standard rotary position embeddings, and that encoding them directly into the position embedding is enough to unlock dense panorama understanding. The paper integrates ERP-RoPE into a vision-language model by reparameterizing each pixel's horizontal position as gamma times f(w), where f(w) is a periodic coordinate peaked at the horizontal center and gamma equals H divided by the sum of cos(theta) over all latitudes, equalizing information density. Trained on the Dense360 dataset, this Dense360VLM-3B improves captioning by 5.92 points and grounding by 16.38 points over the same model without ERP-RoPE, and outperforms a post-trained 4B baseline. The paper also presents Dense360-Bench, with 3,000 grounding and 3,000 captioning questions across 1,279 ERP images balanced across the front, back, left, and right directions, as the first shared evaluation suite for this task.

Load-bearing premise

The whole approach assumes that the automatically generated, reliability-scored captions and masks are faithful ground truth for both training and evaluation, yet the paper reports no human validation of them.

Editorial extensions

If this is right

  • Any multimodal language model that adopts ERP-RoPE can consume a single equirectangular image as input, avoiding the cost of multi-view feature fusion for 360° scenes.
  • Dense360-Bench gives a standard way to compare future models on omnidirectional captioning and grounding, with explicit back-direction evaluation for entities split across the image seam.
  • The reliability-score mechanism in the data pipeline can flag low-confidence caption-mask pairs at scale, making automatic dense annotation more trustworthy.
  • Training with all three data components together yields the best results, suggesting that captioning, referring, and scene-description tasks reinforce each other rather than compete.
  • The back-direction gain supports the specific claim that wrap-around encoding repairs a known failure mode: entities straddling the ERP image's left and right borders.

Reading between the lines

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

  • A natural extension this paper does not test is that the same wrap-around position encoding could transfer to 360° video, cylindrical projections, or panoramic depth and range images, where the same two geometric facts hold.
  • Because the benchmark's captioning score is phrase recall judged by a language model, scores may partly reflect the judge's language preferences; a small human-validated subset would separate caption quality from judge style, but the paper reports no such check.
  • The paper's argument that one ERP image beats multi-view inputs at the input level is a claim about representation efficiency; a direct comparison against multi-view methods at matched compute and data would settle it, but the paper does not run that comparison.
  • If the reliability scores are faithful, the verification pipeline itself is a reusable method for hallucination filtering in dense caption generation, independent of the panorama setting.
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. This paper introduces Dense360, a large automatically generated dataset of 160K equirectangular panoramas with 5M entity-level captions, 1M referring expressions, and 100K entity-grounded scene descriptions; it also presents Dense360-Bench, a captioning/grounding benchmark built from 1,279 curated panoramas, and Dense360VLM, a Qwen2.5VL-based model extended with a proposed ERP-RoPE positional encoding. The authors report that Dense360VLM-3B outperforms a post-trained SA2VA-4B on the benchmark (51.78 vs 47.80 captioning; 76.81 vs 74.39 grounding) and that the proposed ERP-RoPE contributes a large grounding gain in ablations. The central claims are that the dataset, benchmark, and position encoding together constitute a significant step for omnidirectional dense understanding.

Significance. If the benchmark and annotations were independently validated, this would be a useful resource: it is one of the first attempts to scale dense entity captioning and grounding to omnidirectional panoramas, and the three-tier generation pipeline is clearly described. The paper's empirical results are internally consistent, and the geometric motivation for ERP-specific position encoding addresses a real problem. However, the significance is conditional on resolving the benchmark validity issue described below: because training targets and evaluation references come from the same unvalidated automatic pipeline, the reported gains may reflect distribution matching rather than genuine dense understanding. The paper also reports no human-validated subset or external validation, so the central claim is not established as written.

major comments (4)
  1. [§3.2 Verification Pipeline] The validity of Dense360-Bench as a measure of dense understanding is not established. Section 3.2 uses InternVL3 to generate both brief and detailed entity captions; Section 3.3 uses GPT-4o to write scene descriptions from those captions; Section 5.1 evaluates captioning by extracting key phrases from the same detailed captions and asking ChatGPT-4o whether a predicted caption mentions them. Thus both the training targets and the evaluation references are produced by the same automatic pipeline. A model that imitates the generator's output distribution will score high even if it hallucinates semantic content, so the reported gains for Dense360VLM could measure agreement with the generator rather than true understanding. No human validation, human agreement study, or independent benchmark is reported. Please provide a human-validated evaluation subset (e.g., human ratings of caption correctness and human-written references) and rerun the headline results on it, or otherwise demonstrate that the generated references agree with human judgments.
  2. [§3.2] The reliability score only verifies localizability, not semantic correctness. The pipeline computes IoU between a SAM mask obtained after InternVL3 grounds a brief caption and the Level-1 entity mask. This shows that the caption can be located in the entity's region, but it does not verify that the caption's attributes, relations, counts, or predicates are true, because hallucinated details inside the correct region can still yield high IoU. Since InternVL3 also generated the caption, high IoU can reflect self-consistency between the captioning and grounding heads of the same model. Therefore the description of the annotations as 'reliability-scored' overstates what is verified, and the benchmark's grounding and captioning references inherit this unresolved issue.
  3. [§4.1] The derivation of ERP-RoPE is not internally coherent. Equation (1) sums cosθ Wγ over an unspecified set of θ values and then divides by W to obtain γ = H/Σ cosθ, but the reparameterization (h, γf(w)) applies a single global γ while the vertical coordinate g(h)=h is unchanged, so latitude-dependent information density is not actually reflected in per-token encoding. In addition, the implementations of f(w) in Eqs. (3) and (4) do not satisfy the Table 2 properties in a clear way: for W=5, f(1)=1 and f(W)=2, so the periodic extension f(w)=f(w+W) fails at the boundary, and the 'maximum at ERP center' is only approximate. The authors should either correct the derivation, define f(w) on a circle with proper wrap-around boundary conditions, or reframe the contribution as an empirical heuristic with ablations against simpler alternatives such as circular padding of the visual feature map.
  4. [§5.1, §5.2] The grounding evaluation protocol and the ablation interpretation need more support. In Section 5.1, Qwen2.5VL and InternVL3 are evaluated by prompting them to output textual bounding boxes and then running SAM on those boxes, while SA2VA and Dense360VLM directly output masks; this protocol mismatch can confound the comparisons, especially for the grounding metric. In Table 4, the full-data row without ERP-RoPE gives 60.43 grounding accuracy while the full-data row with ERP-RoPE gives 76.81, a 16.38-point jump that is unusually large; no failure-case analysis or comparison with alternative periodic position encodings is provided. I recommend reporting grounding under a common protocol (e.g., all models output masks, or all models output boxes followed by SAM) and adding ablations on a human-validated subset to rule out artifacts of the evaluation pipeline.
minor comments (5)
  1. [§3] The section heading reads 'Dens360 Dataset and Benchmark' and should be corrected to 'Dense360'.
  2. [§5.2] The text says 'as shown in Tab. 3' but the ablation study with data components and ERP-RoPE is reported in Table 4; the cross-reference is wrong.
  3. [Table 3] The table footnote defines the dagger symbol for post-trained MLLMs, but ChatGPT4o-latest has missing entries for all grounding columns without explanation; please add a note that grounding was not evaluated for that model.
  4. [References] Many reference entries contain missing spaces in venue names, e.g., 'InNeurIPS' and 'InCVPR'; please correct the bibliography formatting.
  5. [Limitations] The Limitations paragraph contains a typo, 'a'l the proposed', and the sentence structure should be revised for clarity.

Circularity Check

2 steps flagged · score 6.0 of 10

Dense360-Bench's captioning references are key phrases from the same InternVL3-generated detailed captions used as Dense360VLM training targets, so high captioning scores partly measure imitation of the generator; the reliability score is self-validation by InternVL3, and no human validation anchors the benchmark.

  1. self definitional [Section 3, 'Building Dense360-Bench for Grounding and Captioning' (captioning evaluation protocol)]
    "For captioning tasks, we design a cost-effective evaluation scheme. As shown in Fig. 4, we extract key phrases from each entity’s detailed caption. Given a predicted caption, we formulate a series of yes/no questions to determine whether each key phrase is explicitly mentioned. We employ ChatGPT-4o as the judge model."

    The 'detailed captions' used to define the captioning ground truth are produced by the Level-2 caption pipeline with InternVL3 (Section 3.2), and the same Dense360 dataset (including these detailed captions) is used to fine-tune Dense360VLM (Section 5, 'Baseline Model and Training Datasets'). Thus the captioning metric is recall of phrases written by the same generator that produced the training targets: a model that imitates InternVL3's caption style and hallucination patterns scores high by construction. The benchmark measures agreement with the generator, not verified scene facts; no human validation or agreement study is reported.

  2. other [Section 3.2, 'Verification Pipeline']
    "We note that the grounded SAM mask should be inherently aligned with the input caption, regardless of whether the caption accurately describes the intended entity. This alignment allows us to assess the consistency between the brief caption and the target entity by comparing the SAM-generated mask with the ground-truth entity mask."

    InternVL3 generates the brief captions in the caption pipeline and is then reused as the grounding model in the verification pipeline. The IoU reliability score therefore certifies only that a caption is localizable by the same model that wrote it; it does not verify semantic content such as attributes, counts, or predicates. These reliability-scored captions are subsequently used as training targets for Dense360VLM and as reference phrases in Dense360-Bench, so the benchmark's 'ground truth' inherits the generator's own output. This is a self-validation loop rather than independent verification.

full rationale

The central claim that Dense360VLM achieves 'significant improvements for omnidirectional dense understanding' rests on Dense360-Bench as the measure. For captioning, the benchmark reference is constructed by extracting key phrases from the same InternVL3-generated detailed captions that serve as Dense360VLM's training targets, and the judge is ChatGPT-4o checking phrase presence; this makes high captioning scores partly a measure of imitation of the generator. For the reliability score, InternVL3 both writes the brief captions and grounds them, and the paper itself notes the SAM mask is inherently aligned with the input caption regardless of semantic truth, so the filter does not remove hallucinated content. The grounding benchmark is somewhat less circular because entity masks come from CropFormer and SAM rather than from the caption generator, and the ablations compare models post-trained on the same data, so the relative ERP-RoPE gains have independent content. However, no human validation of the auto-generated annotations is reported, and both training and evaluation draw from the same three-tier pipeline. This is partial circularity, not a fully forced derivation: the positional encoding and grounding comparisons are not reduced to the benchmark by construction. Score 6.

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

The central claims rest on several hand-chosen dataset-generation and encoding parameters plus the assumption that model-generated captions are valid ground truth. No independent physical or formal grounding is added beyond the standard geometry of equirectangular projection.

free parameters (5)
  • gamma (gamma) = gamma = H / sum_theta cos(theta) from Eq. 2
    Introduced to satisfy the asserted constraint that scaled latitude-circle lengths sum to H x W (Eq. 1); the constraint itself is not independently justified.
  • f(w) triangular schedule = Triangular sequence [1, ..., peak, ..., 2] for horizontal position, Eqs. 3 and 4
    Hand-picked 'naive solution' not derived from the properties in Table 2, and no ablation against alternative schedules is provided.
  • Slice overlap stride = 50%
    Choice of overlapping square slice views for entity segmentation in Level-1; affects mask quality and spatial continuity.
  • Mask merging IoU threshold = 0.7
    Entity mask pairs with IoU above 0.7 are merged; the threshold is chosen by hand and affects final mask statistics.
  • Proportion of LLaVA SFT data in training = 30%
    Used in Dense360VLM training alongside Dense360; a training-config hyperparameter that affects reported gains.
assumptions (5)
  • domain assumption MLLM-generated captions and scene descriptions are reliable enough to serve as ground truth for dataset and benchmark.
    Used throughout Sections 3.2 and 3.3 and in Dense360-Bench; no human validation or inter-annotator agreement is reported.
  • ad hoc to paper Eq. 1's normalization constraint (sum of scaled latitude-circle lengths equals H x W) is the right objective for ERP positional encoding.
    Introduced in Section 4.1 without derivation from first principles; the correct spacing for RoPE is not established.
  • ad hoc to paper The four geometric properties in Table 2 are necessary and sufficient for good ERP position encoding.
    Asserted in Section 4.1; no proof that satisfying them yields better MLLM spatial understanding.
  • domain assumption Existing models (CropFormer, RAM++, APE, InternVL3, GPT-4o, SAM, Qwen2.5VL) work as expected in the automatic annotation pipeline.
    The whole dataset relies on these external tools; no quality audit of their outputs beyond model-based reliability scores is reported.
  • domain assumption ChatGPT-4o as judge reliably determines whether key phrases are explicitly mentioned in predicted captions.
    The captioning metric in Dense360-Bench depends on this (Section 3); judge bias is unmeasured.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Dense360: Dense Understanding from Omnidirectional Panoramas." pith.science (2026). https://pith.science/paper/2TIXMAIH

@misc{pith2026250614471,
  author       = {Pith},
  title        = {Pith review of: Dense360: Dense Understanding from Omnidirectional Panoramas},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2TIXMAIH}},
  note         = {Machine review of arXiv:2506.14471}
}
read the original abstract

Multimodal Large Language Models (MLLMs) require comprehensive visual inputs to achieve dense understanding of the physical world. While existing MLLMs demonstrate impressive world understanding capabilities through limited field-of-view (FOV) visual inputs (e.g., 70 degree), we take the first step toward dense understanding from omnidirectional panoramas. We first introduce an omnidirectional panoramas dataset featuring a comprehensive suite of reliability-scored annotations. Specifically, our dataset contains 160K panoramas with 5M dense entity-level captions, 1M unique referring expressions, and 100K entity-grounded panoramic scene descriptions. Compared to multi-view alternatives, panoramas can provide more complete, compact, and continuous scene representations through equirectangular projections (ERP). However, the use of ERP introduces two key challenges for MLLMs: i) spatial continuity along the circle of latitude, and ii) latitude-dependent variation in information density. We address these challenges through ERP-RoPE, a position encoding scheme specifically designed for panoramic ERP. In addition, we introduce Dense360-Bench, the first benchmark for evaluating MLLMs on omnidirectional captioning and grounding, establishing a comprehensive framework for advancing dense visual-language understanding in panoramic settings.

Figures

Figures reproduced from arXiv: 2506.14471 by the authors.

Figure 1
Figure 1. An Example from the Dense360 Dataset. We employ equirectangular projection (ERP) to [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Dataset Generation Pipeline. The Level-1 pipeline generates entity masks. The Level-2 [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Dataset Statistics. We employ Qwen2.5VL-72B-Instruct [ [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: An example for evaluating captioning capability. The caption model refers to the MLLM [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: The architecture of Dense360VLM. The left side illustrates the relationship between [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: More Examples from the Dense360 Dataset. We employ equirectangular projection (ERP) [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 7
Figure 7. Figure 7: More Examples from the Dense360 Dataset. We employ equirectangular projection (ERP) [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]

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. Panoramic Scene Understanding: A Survey from Distortion-Aware Engineering to Sphere-Native Modeling

    cs.CV 2026-06 unverdicted novelty 5.0 of 10

    Survey organizing panoramic scene analysis literature by architectural design and training paradigm, identifying the absence of methods achieving both strict spherical equivariance and full reuse of perspective-pretra...

Reference graph

Works this paper leans on

79 extracted references · 41 canonical work pages · cited by 1 Pith paper

  1. [1]

    Flamingo: a visual language model for few-shot learning

    Jean-Baptiste Alayrac, Jeff Donahue, Pauline Luc, Antoine Miech, Iain Barr, Yana Hasson, Karel Lenc, Arthur Mensch, Katherine Millican, Malcolm Reynolds, et al. Flamingo: a visual language model for few-shot learning. InNeurIPS, 2022. 3

    2022

  2. [2]

    Shuai Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, Sibo Song, Kai Dang, Peng Wang, Shijie Wang, Jun Tang, et al. Qwen2. 5-vl technical report.arXiv preprint arXiv:2502.13923, 2025. 1, 2, 3, 4, 8, 9

    arXiv 2025

  3. [3]

    Egok360: A 360 egocentric kinetic human activity video dataset

    Keshav Bhandari, Mario A DeLaGarza, Ziliang Zong, Hugo Latapie, and Yan Yan. Egok360: A 360 egocentric kinetic human activity video dataset. InICIP, 2020. 3

    work page 2020

  4. [4]

    Language models are few-shot learners

    Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. Language models are few-shot learners. InNeurIPS, 2020. 3

    work page 2020

  5. [5]

    Vip-llava: Making large multimodal models understand arbitrary visual prompts

    Mu Cai, Haotian Liu, Siva Karthik Mustikovela, Gregory P Meyer, Yuning Chai, Dennis Park, and Yong Jae Lee. Vip-llava: Making large multimodal models understand arbitrary visual prompts. InCVPR, 2024. 3

    work page 2024

  6. [6]

    Opening the vocabulary of egocentric actions

    Dibyadip Chatterjee, Fadime Sener, Shugao Ma, and Angela Yao. Opening the vocabulary of egocentric actions. InNeurIPS, 2023. 3

    work page 2023

  7. [7]

    360+ x: A panoptic multi-modal scene understanding dataset

    Hao Chen, Yuqi Hou, Chenyuan Qu, Irene Testini, Xiaohan Hong, and Jianbo Jiao. 360+ x: A panoptic multi-modal scene understanding dataset. InCVPR, 2024. 3

    work page 2024

  8. [8]

    Sharegpt4v: Improving large multi-modal models with better captions

    Lin Chen, Jinsong Li, Xiaoyi Dong, Pan Zhang, Conghui He, Jiaqi Wang, Feng Zhao, and Dahua Lin. Sharegpt4v: Improving large multi-modal models with better captions. InECCV, 2024. 3

    work page 2024

Show all 79 references
  1. [9]

    A single transformer for scalable vision-language modeling

    Yangyi Chen, Xingyao Wang, Hao Peng, and Heng Ji. A single transformer for scalable vision-language modeling. InTMLR, 2024. 3

    2024

  2. [10]

    Expanding performance boundaries of open-source multimodal models with model, data, and test-time scaling.arXiv preprint arXiv:2412.05271, 2024

    Zhe Chen, Weiyun Wang, Yue Cao, Yangzhou Liu, Zhangwei Gao, Erfei Cui, Jinguo Zhu, Shenglong Ye, Hao Tian, Zhaoyang Liu, et al. Expanding performance boundaries of open-source multimodal models with model, data, and test-time scaling.arXiv preprint arXiv:2412.05271, 2024. 8

    2024 arXiv

  3. [11]

    Internvl: Scaling up vision foundation models and aligning for generic visual-linguistic tasks

    Zhe Chen, Jiannan Wu, Wenhai Wang, Weijie Su, Guo Chen, Sen Xing, Muyan Zhong, Qinglong Zhang, Xizhou Zhu, Lewei Lu, et al. Internvl: Scaling up vision foundation models and aligning for generic visual-linguistic tasks. InCVPR, 2024. 3

    2024

  4. [12]

    Embodied artificial intelligence

    Ron Chrisley. Embodied artificial intelligence. InArtificial intelligence, 2003. 1

    2003

  5. [13]

    Xtuner: A toolkit for efficiently fine-tuning llm.https://github.com/InternLM/ xtuner, 2023

    XTuner Contributors. Xtuner: A toolkit for efficiently fine-tuning llm.https://github.com/InternLM/ xtuner, 2023. 8

    2023

  6. [14]

    Instructblip: Towards general-purpose vision-language models with instruction tuning.arXiv preprint arXiv:2305.06500, 2023

    Wenliang Dai, Junnan Li, Dongxu Li, Anthony Meng Huat Tiong, Junqi Zhao, Weisheng Wang, Boyang Li, Pascale Fung, and Steven Hoi. Instructblip: Towards general-purpose vision-language models with instruction tuning.arXiv preprint arXiv:2305.06500, 2023. 3

    2023 arXiv

  7. [15]

    Bert: Pre-training of deep bidirec- tional transformers for language understanding

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. Bert: Pre-training of deep bidirec- tional transformers for language understanding. InNAACL, 2019. 3

    2019

  8. [16]

    Unveiling encoder-free vision-language models.arXiv preprint arXiv:2406.11832, 2024

    Haiwen Diao, Yufeng Cui, Xiaotong Li, Yueze Wang, Huchuan Lu, and Xinlong Wang. Unveiling encoder-free vision-language models.arXiv preprint arXiv:2406.11832, 2024. 3

    2024 arXiv

  9. [17]

    Evev2: Improved baselines for encoder-free vision-language models.arXiv preprint arXiv:2502.06788, 2025

    Haiwen Diao, Xiaotong Li, Yufeng Cui, Yueze Wang, Haoge Deng, Ting Pan, Wenxuan Wang, Huchuan Lu, and Xinlong Wang. Evev2: Improved baselines for encoder-free vision-language models.arXiv preprint arXiv:2502.06788, 2025. 3

    2025 arXiv

  10. [18]

    Pvuw 2025 challenge report: Advances in pixel-level understanding of complex videos in the wild.arXiv preprint arXiv:2504.11326, 2025

    Henghui Ding, Chang Liu, Nikhila Ravi, Shuting He, Yunchao Wei, Song Bai, Philip Torr, Kehuan Song, Xinglin Xie, Kexin Zhang, et al. Pvuw 2025 challenge report: Advances in pixel-level understanding of complex videos in the wild.arXiv preprint arXiv:2504.11326, 2025. 1

    2025 arXiv

  11. [19]

    An image is worth 16x16 words: Transformers for image recognition at scale.arXiv preprint arXiv:2010.11929, 2020

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, et al. An image is worth 16x16 words: Transformers for image recognition at scale.arXiv preprint arXiv:...

    2010 arXiv

  12. [20]

    Embodied videoagent: Persistent memory from egocentric videos and embodied sensors enables dynamic scene understanding

    Yue Fan, Xiaojian Ma, Rongpeng Su, Jun Guo, Rujie Wu, Xi Chen, and Qing Li. Embodied videoagent: Persistent memory from egocentric videos and embodied sensors enables dynamic scene understanding. arXiv preprint arXiv:2501.00358, 2024. 3

    2024 arXiv

  13. [21]

    On path to multimodal generalist: General-level and general-bench

    Hao Fei, Yuan Zhou, Juncheng Li, Xiangtai Li, Qingshan Xu, Bobo Li, Shengqiong Wu, Yaoting Wang, Junbao Zhou, Jiahao Meng, et al. On path to multimodal generalist: General-level and general-bench. arXiv preprint arXiv:2505.04620, 2025. 3

    2025 arXiv

  14. [22]

    Scene-llm: Extending language model for 3d visual reasoning

    Rao Fu, Jingyu Liu, Xilun Chen, Yixin Nie, and Wenhan Xiong. Scene-llm: Extending language model for 3d visual reasoning. InWACV, 2025. 3

    2025

  15. [23]

    Free video-llm: Prompt-guided visual perception for efficient training-free video llms.arXiv preprint arXiv:2410.10441, 2024

    Kai Han, Jianyuan Guo, Yehui Tang, Wei He, Enhua Wu, and Yunhe Wang. Free video-llm: Prompt-guided visual perception for efficient training-free video llms.arXiv preprint arXiv:2410.10441, 2024. 1

    2024 arXiv

  16. [24]

    Lora: Low-rank adaptation of large language models.arXiv preprint arXiv:2106.09685,

    Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. Lora: Low-rank adaptation of large language models.arXiv preprint arXiv:2106.09685,

    arXiv

  17. [25]

    Open-set image tagging with multi-grained text supervision.arXiv preprint arXiv:2310.15200, 2023

    Xinyu Huang, Yi-Jie Huang, Youcai Zhang, Weiwei Tian, Rui Feng, Yuejie Zhang, Yanchun Xie, Yaqian Li, and Lei Zhang. Open-set image tagging with multi-grained text supervision.arXiv preprint arXiv:2310.15200, 2023. 5

    2023 arXiv

  18. [26]

    An egocentric vision-language model based portable real-time smart assistant

    Yifei Huang, Jilan Xu, Baoqi Pei, Yuping He, Guo Chen, Mingfang Zhang, Lijin Yang, Zheng Nie, Jinyao Liu, Guoshun Fan, et al. An egocentric vision-language model based portable real-time smart assistant. arXiv preprint arXiv:2503.04250, 2025. 3 12

    2025 arXiv

  19. [27]

    Scaling up visual and vision-language representation learning with noisy text supervision

    Chao Jia, Yinfei Yang, Ye Xia, Yi-Ting Chen, Zarana Parekh, Hieu Pham, Quoc Le, Yun-Hsuan Sung, Zhen Li, and Tom Duerig. Scaling up visual and vision-language representation learning with noisy text supervision. InICML, 2021. 3

    2021

  20. [28]

    Probres: Probabilistic jump diffusion for open-world egocentric activity recognition.arXiv preprint arXiv:2504.03948, 2025

    Sanjoy Kundu, Shanmukha Vellamchetti, and Sathyanarayanan N Aakur. Probres: Probabilistic jump diffusion for open-world egocentric activity recognition.arXiv preprint arXiv:2504.03948, 2025. 3

    2025

  21. [29]

    Lisa: Reasoning segmentation via large language model

    Xin Lai, Zhuotao Tian, Yukang Chen, Yanwei Li, Yuhui Yuan, Shu Liu, and Jiaya Jia. Lisa: Reasoning segmentation via large language model. InCVPR, 2024. 3, 8

    2024

  22. [30]

    Jrdb-panotrack: An open-world panoptic segmentation and tracking robotic dataset in crowded human environments

    Duy Tho Le, Chenhui Gou, Stavya Datta, Hengcan Shi, Ian Reid, Jianfei Cai, and Hamid Rezatofighi. Jrdb-panotrack: An open-world panoptic segmentation and tracking robotic dataset in crowded human environments. InCVPR, 2024. 3

    2024

  23. [31]

    Perspective-aware reasoning in vision-language models via mental imagery simulation.arXiv preprint arXiv:2504.17207, 2025

    Phillip Y Lee, Jihyeon Je, Chanho Park, Mikaela Angelina Uy, Leonidas Guibas, and Minhyuk Sung. Perspective-aware reasoning in vision-language models via mental imagery simulation.arXiv preprint arXiv:2504.17207, 2025. 3

    2025 arXiv

  24. [32]

    360 vision, from panoramas to vr

    Laurent Lescop. 360 vision, from panoramas to vr. InEnvisioning architecture: space/time/meaning, 2017. 1

    2017

  25. [33]

    Llava-onevision: Easy visual task transfer.arXiv preprint arXiv:2408.03326,

    Bo Li, Yuanhan Zhang, Dong Guo, Renrui Zhang, Feng Li, Hao Zhang, Kaichen Zhang, Peiyuan Zhang, Yanwei Li, Ziwei Liu, et al. Llava-onevision: Easy visual task transfer.arXiv preprint arXiv:2408.03326,

    arXiv

  26. [34]

    Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models

    Junnan Li, Dongxu Li, Silvio Savarese, and Steven Hoi. Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models. InICML, 2023. 3

    2023

  27. [35]

    Blip: Bootstrapping language-image pre-training for unified vision-language understanding and generation

    Junnan Li, Dongxu Li, Caiming Xiong, and Steven Hoi. Blip: Bootstrapping language-image pre-training for unified vision-language understanding and generation. InICML, 2022. 3

    2022

  28. [36]

    Bevformer: learning bird’s-eye-view representation from lidar-camera via spatiotemporal transformers

    Zhiqi Li, Wenhai Wang, Hongyang Li, Enze Xie, Chonghao Sima, Tong Lu, Qiao Yu, and Jifeng Dai. Bevformer: learning bird’s-eye-view representation from lidar-camera via spatiotemporal transformers. In TPAMI, 2024. 1

    2024

  29. [37]

    Describe anything: Detailed localized image and video captioning.arXiv preprint arXiv:2504.16072, 2025

    Long Lian, Yifan Ding, Yunhao Ge, Sifei Liu, Hanzi Mao, Boyi Li, Marco Pavone, Ming-Yu Liu, Trevor Darrell, Adam Yala, et al. Describe anything: Detailed localized image and video captioning.arXiv preprint arXiv:2504.16072, 2025. 3

    2025 arXiv

  30. [38]

    Kitti-360: A novel dataset and benchmarks for urban scene understanding in 2d and 3d

    Yiyi Liao, Jun Xie, and Andreas Geiger. Kitti-360: A novel dataset and benchmarks for urban scene understanding in 2d and 3d. InTPAMI, 2022. 3

    2022

  31. [39]

    Ureca: Unique region caption anything.arXiv preprint arXiv:2504.05305, 2025

    Sangbeom Lim, Junwan Kim, Heeji Yoon, Jaewoo Jung, and Seungryong Kim. Ureca: Unique region caption anything.arXiv preprint arXiv:2504.05305, 2025. 3

    2025 arXiv

  32. [40]

    Egocentric video-language pretraining

    Kevin Qinghong Lin, Jinpeng Wang, Mattia Soldan, Michael Wray, Rui Yan, Eric Z Xu, Difei Gao, Rong-Cheng Tu, Wenzhe Zhao, Weijie Kong, et al. Egocentric video-language pretraining. InNeurIPS,

  33. [41]

    Improved baselines with visual instruction tuning.arXiv preprint arXiv:2310.03744, 2024

    Haotian Liu, Chunyuan Li, Yuheng Li, and Yong Jae Lee. Improved baselines with visual instruction tuning.arXiv preprint arXiv:2310.03744, 2024. 1, 3, 8

    2024 arXiv

  34. [42]

    Improved baselines with visual instruction tuning

    Haotian Liu, Chunyuan Li, Yuheng Li, and Yong Jae Lee. Improved baselines with visual instruction tuning. InCVPR, pages 26296–26306, 2024. 3

    2024

  35. [43]

    Visual instruction tuning.arXiv preprint arXiv:2304.08485, 2023

    Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. Visual instruction tuning.arXiv preprint arXiv:2304.08485, 2023. 3

    2023 arXiv

  36. [44]

    Visual instruction tuning

    Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. Visual instruction tuning. InNeurIPS, 2023. 3

    2023

  37. [45]

    Seg-zero: Reasoning- chain guided segmentation via cognitive reinforcement.arXiv preprint arXiv:2503.06520, 2025

    Yuqi Liu, Bohao Peng, Zhisheng Zhong, Zihao Yue, Fanbin Lu, Bei Yu, and Jiaya Jia. Seg-zero: Reasoning- chain guided segmentation via cognitive reinforcement.arXiv preprint arXiv:2503.06520, 2025. 3

    2025 arXiv

  38. [46]

    Vmamba: Visual state space model

    Yue Liu, Yunjie Tian, Yuzhong Zhao, Hongtian Yu, Lingxi Xie, Yaowei Wang, Qixiang Ye, Jianbin Jiao, and Yunfan Liu. Vmamba: Visual state space model. InNeurIPS, 2024. 3

    2024

  39. [47]

    Mono-internvl: Pushing the boundaries of monolithic multimodal large language models with endogenous visual pre-training

    Gen Luo, Xue Yang, Wenhan Dou, Zhaokai Wang, Jiawen Liu, Jifeng Dai, Yu Qiao, and Xizhou Zhu. Mono-internvl: Pushing the boundaries of monolithic multimodal large language models with endogenous visual pre-training. InCVPR, 2025. 3

    2025

  40. [48]

    Feast your eyes: Mixture-of-resolution adaptation for multimodal large language models.arXiv preprint arXiv:2403.03003,

    Gen Luo, Yiyi Zhou, Yuxin Zhang, Xiawu Zheng, Xiaoshuai Sun, and Rongrong Ji. Feast your eyes: Mixture-of-resolution adaptation for multimodal large language models.arXiv preprint arXiv:2403.03003,

    arXiv

  41. [49]

    Wmnav: Integrating vision-language models into world models for object goal navigation.arXiv preprint arXiv:2503.02247, 2025

    Dujun Nie, Xianda Guo, Yiqun Duan, Ruijun Zhang, and Long Chen. Wmnav: Integrating vision-language models into world models for object goal navigation.arXiv preprint arXiv:2503.02247, 2025. 3

    2025 arXiv

  42. [50]

    An introduction to convolutional neural networks.arXiv preprint arXiv:1511.08458, 2015

    Keiron O’shea and Ryan Nash. An introduction to convolutional neural networks.arXiv preprint arXiv:1511.08458, 2015. 3

    2015 arXiv

  43. [51]

    High quality entity segmentation

    Lu Qi, Jason Kuen, Tiancheng Shen, Jiuxiang Gu, Wenbo Li, Weidong Guo, Jiaya Jia, Zhe Lin, and Ming-Hsuan Yang. High quality entity segmentation. InICCV, 2023. 5

    2023

  44. [52]

    Learning transferable visual models from natural language supervision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learning transferable visual models from natural language supervision. InICML, 2021. 3

    2021

  45. [53]

    Eve: Efficient multimodal vision language models with elastic visual experts.arXiv preprint arXiv:2501.04322, 2025

    Miao Rang, Zhenni Bi, Chuanjian Liu, Yehui Tang, Kai Han, and Yunhe Wang. Eve: Efficient multimodal vision language models with elastic visual experts.arXiv preprint arXiv:2501.04322, 2025. 1

    2025 arXiv

  46. [54]

    Sam 2: Segment anything in images and videos.arXiv preprint arXiv:2408.00714, 2024

    Nikhila Ravi, Valentin Gabeur, Yuan-Ting Hu, Ronghang Hu, Chaitanya Ryali, Tengyu Ma, Haitham Khedr, Roman Rädle, Chloe Rolland, Laura Gustafson, et al. Sam 2: Segment anything in images and videos.arXiv preprint arXiv:2408.00714, 2024. 6, 8 13

    2024 arXiv

  47. [55]

    Vlm-r1: A stable and generalizable r1-style large vision-language model.arXiv preprint arXiv:2504.07615, 2025

    Haozhan Shen, Peng Liu, Jingcheng Li, Chunxin Fang, Yibo Ma, Jiajia Liao, Qiaoli Shen, Zilun Zhang, Kangjia Zhao, Qianqian Zhang, et al. Vlm-r1: A stable and generalizable r1-style large vision-language model.arXiv preprint arXiv:2504.07615, 2025. 3

    2025 arXiv

  48. [56]

    Aligning and prompting everything all at once for universal visual perception

    Yunhang Shen, Chaoyou Fu, Peixian Chen, Mengdan Zhang, Ke Li, Xing Sun, Yunsheng Wu, Shaohui Lin, and Rongrong Ji. Aligning and prompting everything all at once for universal visual perception. In CVPR, 2024. 5

    2024

  49. [57]

    Long-vita: Scaling large multi-modal models to 1 million tokens with leading short-context accuray.arXiv preprint arXiv:2502.05177, 2025

    Yunhang Shen, Chaoyou Fu, Shaoqi Dong, Xiong Wang, Peixian Chen, Mengdan Zhang, Haoyu Cao, Ke Li, Xiawu Zheng, Yan Zhang, et al. Long-vita: Scaling large multi-modal models to 1 million tokens with leading short-context accuray.arXiv preprint arXiv:2502.05177, 2025. 1

    2025

  50. [58]

    Llm-seg: Bridging image segmentation and large language model reasoning

    Junchi Wang and Lei Ke. Llm-seg: Bridging image segmentation and large language model reasoning. In CVPR, 2024. 3

    2024

  51. [59]

    Qwen2-vl: Enhancing vision-language model’s perception of the world at any resolution.arXiv preprint arXiv:2409.12191, 2024

    Peng Wang, Shuai Bai, Sinan Tan, Shijie Wang, Zhihao Fan, Jinze Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, et al. Qwen2-vl: Enhancing vision-language model’s perception of the world at any resolution.arXiv preprint arXiv:2409.12191, 2024. 3

    2024 arXiv

  52. [60]

    Controlmllm: Training-free visual prompt learning for multimodal large language models.NeurIPS, 2024

    Mingrui Wu, Xinyue Cai, Jiayi Ji, Jiale Li, Oucheng Huang, Gen Luo, Hao Fei, Guannan Jiang, Xiaoshuai Sun, and Rongrong Ji. Controlmllm: Training-free visual prompt learning for multimodal large language models.NeurIPS, 2024. 1

    2024

  53. [61]

    Panovos: Bridging non-panoramic and panoramic views with transformer for video segmentation

    Shilin Yan, Xiaohao Xu, Renrui Zhang, Lingyi Hong, Wenchao Chen, Wenqiang Zhang, and Wei Zhang. Panovos: Bridging non-panoramic and panoramic views with transformer for video segmentation. In ECCV, 2024. 3

    2024

  54. [62]

    An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, et al. Qwen2. 5 technical report.arXiv preprint arXiv:2412.15115, 2024. 4

    2024 arXiv

  55. [63]

    The dawn of lmms: Preliminary explorations with gpt-4v (ision).arXiv preprint arXiv:2309.17421, 2023

    Zhengyuan Yang, Linjie Li, Kevin Lin, Jianfeng Wang, Chung-Ching Lin, Zicheng Liu, and Lijuan Wang. The dawn of lmms: Preliminary explorations with gpt-4v (ision).arXiv preprint arXiv:2309.17421, 2023. 3

    2023 arXiv

  56. [64]

    Lavt: Language- aware vision transformer for referring image segmentation

    Zhao Yang, Jiaqi Wang, Yansong Tang, Kai Chen, Hengshuang Zhao, and Philip HS Torr. Lavt: Language- aware vision transformer for referring image segmentation. InCVPR, 2022. 3

    2022

  57. [65]

    Sa2va: Marrying sam2 with llava for dense grounded understanding of images and videos.arXiv preprint arXiv:2501.04001, 2025

    Haobo Yuan, Xiangtai Li, Tao Zhang, Zilong Huang, Shilin Xu, Shunping Ji, Yunhai Tong, Lu Qi, Jiashi Feng, and Ming-Hsuan Yang. Sa2va: Marrying sam2 with llava for dense grounded understanding of images and videos.arXiv preprint arXiv:2501.04001, 2025. 1, 8, 9

    2025 arXiv

  58. [66]

    4th pvuw mevis 3rd place report: Sa2va.arXiv preprint arXiv:2504.00476, 2025

    Haobo Yuan, Tao Zhang, Xiangtai Li, Lu Qi, Zilong Huang, Shilin Xu, Jiashi Feng, and Ming-Hsuan Yang. 4th pvuw mevis 3rd place report: Sa2va.arXiv preprint arXiv:2504.00476, 2025. 1

    2025 arXiv

  59. [67]

    A survey of autonomous driving: Common practices and emerging technologies

    Ekim Yurtsever, Jacob Lambert, Alexander Carballo, and Kazuya Takeda. A survey of autonomous driving: Common practices and emerging technologies. InIEEE access, 2020. 1

    2020

  60. [68]

    Clip2: Contrastive language-image-point pretraining from real-world point cloud data

    Yihan Zeng, Chenhan Jiang, Jiageng Mao, Jianhua Han, Chaoqiang Ye, Qingqiu Huang, Dit-Yan Yeung, Zhen Yang, Xiaodan Liang, and Hang Xu. Clip2: Contrastive language-image-point pretraining from real-world point cloud data. InCVPR, 2023. 3

    2023

  61. [69]

    Mem2ego: Empowering vision-language models with global-to-ego memory for long-horizon embodied navigation

    Lingfeng Zhang, Yuecheng Liu, Zhanguang Zhang, Matin Aghaei, Yaochen Hu, Hongjian Gu, Moham- mad Ali Alomrani, David Gamaliel Arcos Bravo, Raika Karimi, Atia Hamidizadeh, et al. Mem2ego: Empowering vision-language models with global-to-ego memory for long-horizon embodied navi...

    2025 arXiv

  62. [70]

    Omg-llava: Bridging image-level, object-level, pixel-level reasoning and understanding

    Tao Zhang, Xiangtai Li, Hao Fei, Haobo Yuan, Shengqiong Wu, Shunping Ji, Chen Change Loy, and Shuicheng Yan. Omg-llava: Bridging image-level, object-level, pixel-level reasoning and understanding. InNeurIPS, 2024. 1

    2024

  63. [71]

    Pixel-sail: Single transformer for pixel-grounded understanding.arXiv preprint arXiv:2504.10465, 2025

    Tao Zhang, Xiangtai Li, Zilong Huang, Yanwei Li, Weixian Lei, Xueqing Deng, Shihao Chen, Shunping Ji, and Jiashi Feng. Pixel-sail: Single transformer for pixel-grounded understanding.arXiv preprint arXiv:2504.10465, 2025. 1, 3, 8

    2025 arXiv

  64. [72]

    Dvis: Decoupled video instance segmentation framework

    Tao Zhang, Xingye Tian, Yu Wu, Shunping Ji, Xuebo Wang, Yuan Zhang, and Pengfei Wan. Dvis: Decoupled video instance segmentation framework. InICCV, 2023. 1

    2023

  65. [73]

    Dvis++: Improved decoupled framework for universal video segmentation

    Tao Zhang, Xingye Tian, Yikang Zhou, Shunping Ji, Xuebo Wang, Xin Tao, Yuan Zhang, Pengfei Wan, Zhongyuan Wang, and Yu Wu. Dvis++: Improved decoupled framework for universal video segmentation. IEEE TPAMI, 2025. 1

    2025

  66. [74]

    Seeing clearly by layer two: Enhancing attention heads to alleviate hallucination in lvlms.arXiv preprint arXiv:2411.09968, 2024

    Xiaofeng Zhang, Yihao Quan, Chaochen Gu, Chen Shen, Xiaosong Yuan, Shaotian Yan, Hao Cheng, Kaijie Wu, and Jieping Ye. Seeing clearly by layer two: Enhancing attention heads to alleviate hallucination in lvlms.arXiv preprint arXiv:2411.09968, 2024. 1

    2024 arXiv

  67. [75]

    Enhancing multimodal large language models complex reason via similarity computation

    Xiaofeng Zhang, Fanshuo Zeng, Yihao Quan, Zheng Hui, and Jiawei Yao. Enhancing multimodal large language models complex reason via similarity computation. InAAAI, 2025. 1

    2025

  68. [76]

    Regionclip: Region-based language-image pretraining

    Yiwu Zhong, Jianwei Yang, Pengchuan Zhang, Chunyuan Li, Noel Codella, Liunian Harold Li, Luowei Zhou, Xiyang Dai, Lu Yuan, Yin Li, et al. Regionclip: Region-based language-image pretraining. In CVPR, 2022. 3

    2022

  69. [77]

    Improving video segmentation via dynamic anchor queries

    Yikang Zhou, Tao Zhang, Shunping Ji, Shuicheng Yan, and Xiangtai Li. Improving video segmentation via dynamic anchor queries. InECCV, 2024. 1

    2024

  70. [78]

    Are they the same? exploring visual correspondence shortcomings of multimodal llms.arXiv preprint arXiv:2501.04670, 2025

    Yikang Zhou, Tao Zhang, Shilin Xu, Shihao Chen, Qianyu Zhou, Yunhai Tong, Shunping Ji, Jiangning Zhang, Xiangtai Li, and Lu Qi. Are they the same? exploring visual correspondence shortcomings of multimodal llms.arXiv preprint arXiv:2501.04670, 2025. 1

    2025 arXiv

  71. [79]

    Internvl3: Exploring advanced training and test-time recipes for open-source 14 multimodal models.arXiv preprint arXiv:2504.10479, 2025

    Jinguo Zhu, Weiyun Wang, Zhe Chen, Zhaoyang Liu, Shenglong Ye, Lixin Gu, Yuchen Duan, Hao Tian, Weijie Su, Jie Shao, et al. Internvl3: Exploring advanced training and test-time recipes for open-source 14 multimodal models.arXiv preprint arXiv:2504.10479, 2025. 1, 2, 3, 6, 8, 9 15

    2025 arXiv

Pith tools

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