Pith. sign in

REVIEW 3 major objections 7 minor 2 cited by

TimeRefine: Temporal Grounding with Time Refining Video LLM

T0 review · 3 major / 7 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 TimeRefine claims that reformulating video temporal grounding as iterative offset refinement—rough segment followed by repeated corrections, trained with an L1 auxiliary head—yields 3.6% and 5.0% mIoU gains over VTimeLLM on ActivityNet…

desk verdict Solid, well-ablated engineering contribution to temporal grounding in video LLMs; the “self-correction” story is weaker than the numbers because the training sequence never uses the model’s own predictions. read the letter →

arxiv 2412.09601 v2 pith:SEZXQ77K submitted 2024-12-12 cs.CV cs.AIcs.CL

Xizi Wang , Feng Cheng , Ziyang Wang , Huiyu Wang , Md Mohaiminul Islam , Lorenzo Torresani , Mohit Bansal , Gedas Bertasius
show 1 more author
David Crandall
This is my paper · ORCID
classification cs.CVcs.AIcs.CL
keywords temporalgroundingvideolargelanguagemodelsiterativerefinementoffsetpredictionL1auxiliarylosscoarse-to-fineActivityNetCaptionsCharades-STA
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

TimeRefine argues that Video LLMs fail at temporal grounding largely because the standard training target is mismatched to the regression nature of timestamps: cross-entropy loss treats a prediction of 21s and 100s as equally wrong when the ground truth is 20s, and only two answer tokens carry localization signal. The paper's solution is to reformulate grounding as an iterative refinement task in which the model first predicts a rough segment and then predicts offsets that move that segment toward the target, repeating this correction several times. Training examples are produced by adding Gaussian noise of decreasing variance (schedule $\sigma^2 = \{5,3,1,0\}$ seconds over $K=4$ steps) to ground-truth timestamps, and an auxiliary L1 head on the refinement token supplies distance-aware supervision. With the same training data and settings, TimeRefine reports 3.6% and 5.0% mIoU improvements over VTimeLLM on ActivityNet Captions and Charades-STA, and 1.2% over VTG-LLM on Charades-STA. If correct, the work shows that localization ability in Video LLMs can be bought by changing what the model is asked to predict, rather than by building new visual or temporal encoders.

What carries the argument

The carrying object is the iterative time refinement sequence with its control-token grammar, paired with the auxiliary L1 regression head attached to the <refine> token. The sequence turns one sparse timestamp prediction into several rounds of segment-and-offset predictions, so the model receives dense, distance-aware supervision and practices self-correction; the decreasing noise schedule $\sigma^2=\{5,3,1,0\}$ simulates a coarse-to-fine path from rough guess to exact target. The auxiliary head, a single linear layer on the hidden state of <refine>, provides the gradient that cross-entropy cannot, namely a penalty proportional to how far the predicted segment lies from the ground truth, and it can be removed after training without losing accuracy.

What would settle it

Record the first-round segment predictions TimeRefine makes on a held-out set, compare the empirical distribution of their start/end errors to the $\mathcal{N}(0,\sigma^2)$ training noise used to generate refinement sequences, and check whether refinement gains persist when the two distributions differ substantially; a test set with systematically biased first guesses (for example, all target events placed near one end of the video) would expose a mismatch that the fixed symmetric noise never simulates.

Watch

Extended reading notes

Core claim

The central discovery is that a coarse-to-fine temporal refinement sequence is a better learning target for LLM-based temporal grounding than direct timestamp prediction. For a target segment $S=(s,e)$, the method converts the ground truth into the sequence $S' = ((s_0,e_0,o^s_0,o^e_0),\ldots,(s_K,e_K,o^s_K,o^e_K))$ with $s=s_k+o^s_k$ and $e=e_k+o^e_k$, serialized with control tokens <seg_start>, <offset>, <refine>, and <seg_end>. The rough guesses $s_k,e_k$ are synthesized in training by sampling offsets from zero-mean Gaussians with variances $\{5,3,1,0\}$ seconds at the four steps, so the model is asked to correct perturbations that shrink toward the ground truth. At every <refine> token, a lightweight linear head predicts the current segment and is trained with an L1 loss, while the LLM's ordinary cross-entropy loss is retained. At inference the auxiliary head is discarded and the final prediction is read from the last refinement step, $s_K+o^s_K$ and $e_K+o^e_K$. The paper reports that this scheme improves VTimeLLM by 3.6 mIoU on ActivityNet Captions and 5.0 mIoU on Charades-STA, and VTG-LLM by 1.2 mIoU on Charades-STA.

Load-bearing premise

The load-bearing premise is that Gaussian noise added to ground-truth timestamps during training produces rough predictions whose error pattern matches the model's own initial mistakes at test time, so the offset corrections learned under that synthetic noise will transfer to real errors.

Editorial extensions

If this is right

  • On ActivityNet Captions, adding TimeRefine to VTimeLLM-7B is reported to lift R@0.3 from 44.0 to 48.0, R@0.5 from 27.8 to 33.6, R@0.7 from 14.3 to 17.5, and mIoU from 30.4 to 34.0.
  • On Charades-STA, VTimeLLM-TIMEREFINE is reported to lift mIoU from 31.2 to 36.2, and VTG-LLM-TIMEREFINE from 34.4 to 35.6.
  • The auxiliary head need not exist at test time: decoding from the last refinement step, from the auxiliary head, or from both yields essentially the same mIoU, so the method remains compatible with standard next-token generation.
  • The refinement objective also improves dense video captioning metrics on ActivityNet Captions, suggesting temporal understanding gains transfer to related tasks.
  • Because only the supervised target and loss are changed, the method can be layered onto any LLM-based temporal grounding model without modifying its architecture.

Reading between the lines

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

  • An unstated extension is to replace the fixed Gaussian noise schedule with noise matched to the model's measured first-round error distribution; the paper's ablation already shows fixed noise beats duration-normalized noise, but it never tests error-matched noise.
  • The offset-refinement pattern is a generic remedy for any continuous output an LLM must generate as tokens, so the same trick may transfer to bounding-box regression, audio event onsets, or 3D localization tasks that suffer from the same cross-entropy blindness.
  • A testable prediction is that refinement gains are concentrated on segments whose first rough guess is near but not exactly on target; for very short or highly ambiguous events, extra refinement tokens might add compute without improving IoU.
  • One could also train with fewer refinement steps and, at inference, feed the model's own last output back as the next rough guess, turning the learned correction behavior into an open-ended self-improvement loop.
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 / 7 minor

Summary. This paper proposes TimeRefine, a training-objective modification for Video-LLM temporal grounding. Instead of directly generating start and end timestamps, the model generates a coarse segment and then predicts offsets to the target, repeating this over K refinement steps; an auxiliary L1 head on the <refine> token is added during training and discarded at inference. On ActivityNet Captions and Charades-STA, applying TimeRefine to VTimeLLM improves mIoU from 30.4 to 34.0 and from 31.2 to 36.2, and applying it to VTG-LLM improves Charades-STA mIoU from 34.4 to 35.6. The paper also reports ablations of the refinement target, number of steps, noise schedule, auxiliary loss type, and decoding strategy.

Significance. The empirical contribution is clearly presented and well ablated: the authors isolate offset prediction, K=4 with the {5,3,1,0} schedule, and the L1 auxiliary loss, and they report consistent gains across two base models and datasets under matched training data. The plug-and-play nature of the method and the promise of code release increase its value. The main reservation is interpretational: during training the rough segments are sampled from ground truth plus Gaussian noise, and the model is never conditioned on its own predictions, so the 'self-correction' mechanism is not directly supported. The reported gains may still hold as a coarse-to-fine denoising curriculum with auxiliary supervision, but the paper should test this before the self-correction claim is accepted.

major comments (3)
  1. [Sec. 3.2.2 (Eqs. 3-4)] The central 'self-correct' claim is not supported by the training procedure. The model never sees its own rough prediction during training: (s_k,e_k) are obtained by adding Gaussian noise to the ground-truth timestamps, and the target offsets are exactly the sampled noise. At inference, the model's initial segment comes from its own autoregressive generation, whose error distribution may differ from the zero-mean, isotropic, truncated Gaussian used in training. The reported gains could therefore come from the denoising curriculum and the auxiliary L1 head rather than from learning to correct the model's own errors. Please provide (a) a quantitative comparison of the test-time error distribution of the model's first generated segment with the training noise schedule, and (b) an experiment in which refinement is trained on the model's own generated rough segments (e.g., scheduled sampling or iterative self-training), or, at minimum, reframe the contribution as coarse-to-fine/denoising supervision rather than self-correction.
  2. [Sec. 4.4 (Tables 2-5)] There is no control that isolates sequential refinement from the extra supervision tokens. A variant that predicts K segments in parallel, or predicts a coarse segment and the target without using the offsets as conditioning, would test whether the improvement is due to the refinement mechanism or merely to having more regression targets. The IoU-prediction variant in Table 2, which also improves over no refinement, suggests that additional feedback alone can help. This does not invalidate the empirical gains, but it is necessary to support the paper's interpretation that the model learns to self-correct through iterative refinement.
  3. [Sec. 3.2.2 (Eq. 2) and Sec. 3.4.1] The indexing of the refinement sequence is inconsistent. Equation (2) indexes tuples from 0 to K, giving K+1 tuples, but Eq. (3) defines K Gaussian distributions for k=1..K, and Sec. 3.4.1 sets K=4 with four variance values. The construction of the initial tuple (s0,e0,o^s0,o^e0) is never defined, and it is unclear whether the zero-noise final step is part of the target sequence. Please fix the indexing so that Eq. (3)-(4) and the K=4 schedule are consistent and reproducible.
minor comments (7)
  1. [Sec. 3.2.2, after Eq. (2)] The sentence 's = s_k + o^s_k, e_i = e_k + o^e_k' contains a typo: e_i should be e_k, and the superscripted offset notation in the tuple is inconsistent with the subscripted notation used in the following equations.
  2. [Sec. 3.4.1 and Table 3] The text calls sigma^2 = {5,3,1,0} a set of standard deviations, while Eq. (3) defines N(0,sigma^2_k); please use consistent notation for variances versus standard deviations.
  3. [Figure 1] The example shows only two refinement steps (the second <refine> yields the final 20s to 25s), whereas the default setting is K=4; please label the figure as illustrative or extend it to match the default schedule.
  4. [Eq. (6)] The L1 term 1/(2|S|)|Ŝ - S| should specify the norm and the summation over segments and endpoints; as written, the normalization is ambiguous when |S| > 1.
  5. [Sec. 4.4.1] The sentence 'We apply Gaussian noise with a standard deviation of 5 seconds' for the IoU/offset comparison should state the full schedule {5,0}, matching Table 3's K=2 row; otherwise Tables 2 and 3 are hard to reconcile.
  6. [Sec. 3.3] The phrase 'discarded during inference ˙For the prediction' contains a stray punctuation mark before 'For'.
  7. [Sec. 4.2 and Table 1] The claim of zero-shot evaluation for VTG-LLM-TIMEREFINE on Charades-STA should be stated more carefully, and the paper should clarify whether the VTimeLLM-TIMEREFINE model is evaluated on the same splits used for training, since the zero-shot terminology could be misleading.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: TimeRefine's refinement training is a supervised denoising curriculum with external-benchmark evaluation, not a prediction that reduces to its own inputs.

full rationale

TimeRefine's training sequences are generated by adding Gaussian noise to ground-truth timestamps (Eqs. 3-4), and the offset targets are exactly the sampled noise; the model is optimized with CE and an auxiliary L1 loss against the same ground-truth segments. This is a standard supervised training curriculum, not a circular derivation: no parameter is fitted to the evaluation metric, and the final test-time prediction (last refined segment plus predicted offsets) is not equal to any training input by construction. The reported mIoU gains are on external benchmarks (ActivityNet Captions, Charades-STA) under the same training settings as the VTimeLLM and VTG-LLM baselines, so the central claim is independently falsifiable. The only author-overlap citations (Cheng et al. [2023] VindLU; Cheng and Bertasius [2022] TallFormer) appear in related-work lists and are not load-bearing for the method or its evaluation. The skeptic's train/test mismatch—training refines GT-plus-noise rather than the model's own predicted segments—is a legitimate generalization risk, but it is a correctness/robustness concern, not a circularity: the test prediction does not reduce to a fitted parameter or to a self-citation. Therefore no circular step is present.

Assumptions & free parameters 4 free parameters · 3 assumptions · 0 invented entities

The method introduces no new physical entities. Its free parameters (K, sigma schedule, lambda, LoRA config) are tuned via ablations on validation data. The core assumptions are domain assumptions about the distribution of prediction errors and the behavior of cross-entropy on timestamp tokens, both of which are reasonable but not empirically verified in the paper.

free parameters (4)
  • number of refinement steps K = 4
    Ablated in Table 3; K=4 gives the best mIoU on ActivityNet.
  • Gaussian noise schedule sigma^2 = {5,3,1,0} seconds
    Ablated in Table 3; a fixed decreasing schedule is chosen over adaptive schedules.
  • L1 loss weight lambda = 10
    Set in Section 3.4.1 as a hyperparameter to balance CE and L1 losses; no ablation curve shown for lambda.
  • LoRA rank and alpha = 64 and 128
    From Section 3.4.2, the LoRA configuration used for VTimeLLM integration.
assumptions (3)
  • domain assumption Gaussian noise added to ground truth timestamps simulates the model's initial prediction errors.
    Section 3.2.2, Eq. (3): offsets are sampled from N(0, sigma^2) to create rough predictions; the transfer to test time assumes this distribution matches the model's self-generated errors.
  • domain assumption Timestamps tokenized as text tokens lose numeric distance information under cross-entropy loss.
    Section 3.2.3 argues CE gives identical penalty for '21s' and '100s'; this motivates the auxiliary L1 head.
  • domain assumption The hidden state of the <refine> token carries sufficient temporal information for a linear head to predict the segment.
    Section 3.2.3, Eq. (5); validated indirectly by Table 4, which shows the L1 head improves mIoU by 1.6.

how reviews work

0 comments
Cite this review

Pith. "Pith review of TimeRefine: Temporal Grounding with Time Refining Video LLM." pith.science (2026). https://pith.science/paper/SEZXQ77K

@misc{pith2026241209601,
  author       = {Pith},
  title        = {Pith review of: TimeRefine: Temporal Grounding with Time Refining Video LLM},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SEZXQ77K}},
  note         = {Machine review of arXiv:2412.09601}
}
read the original abstract

Video temporal grounding aims to localize relevant temporal boundaries in a video given a textual prompt. Recent work has focused on enabling Video LLMs to perform video temporal grounding via next-token prediction of temporal timestamps. However, accurately localizing timestamps in videos remains challenging for Video LLMs when relying solely on temporal token prediction. Our proposed TimeRefine addresses this challenge in two ways. First, instead of directly predicting the start and end timestamps, we reformulate the temporal grounding task as a temporal refining task: the model first makes rough predictions and then refines them by predicting offsets to the target segment. This refining process is repeated multiple times, through which the model progressively self-improves its temporal localization accuracy. Second, to enhance the model's temporal perception capabilities, we incorporate an auxiliary prediction head that penalizes the model more if a predicted segment deviates further from the ground truth, thus encouraging the model to make closer and more accurate predictions. Our plug-and-play method can be integrated into most LLM-based temporal grounding approaches. The experimental results demonstrate that TimeRefine achieves 3.6% and 5.0% mIoU improvements on the ActivityNet and Charades-STA datasets, respectively. Code and pretrained models will be released.

Figures

Figures reproduced from arXiv: 2412.09601 by the authors.

Figure 1
Figure 1. Given a text query, existing Video LLMs directly predict the start and end timestamps, [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Overview of the TIMEREFINE. Given a video and a textual user prompt, our model predicts an iterative time refinement sequence, i.e., an initial rough estimation of the boundary, followed by new predictions and offsets based on its previous predictions. The new predictions and offsets can help the model learn how to refine its predictions and correct its errors. Our empirical experiments show that such an iterative t… view at source ↗
Figure 3
Figure 3. Zero-shot case study. We compare the output of VTimeLLM, VTG-LLM and TIMERE￾FINE on a video from Charades-STA dataset. Our method iteratively refines the segment predictions. The final prediction achieves an IoU of 0.95, which is the highest among all predictions. Method ActivityNet Captions R@0.3 R@0.5 R@0.7 mIoU No Refinement 44 27.8 14.3 30.4 IoU prediction 44.8 28.3 14.8 30.9 offset prediction 45.2 29.0 15.4 31.… view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

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

  1. TimePLE: Rethinking Temporal Representation for Video Temporal Grounding

    cs.CV 2026-07 conditional novelty 6.0 of 10

    TimePLE predicts a whole video interval as a joint distribution over a position-duration square, rather than predicting start and end separately, and reports higher mIoU across four VTG benchmarks.

  2. A Survey on Video Temporal Grounding with Multimodal Large Language Model

    cs.CV 2025-08 unverdicted novelty 3.0 of 10

    A taxonomized review of video temporal grounding with multimodal large language models, covering model roles, training paradigms, feature processing, benchmarks, and open problems.

Reference graph

Works this paper leans on

25 extracted references · 4 canonical work pages · cited by 2 Pith papers

  1. [8]

    Vtg-llm: Integrating timestamp knowledge into video llms for enhanced video temporal grounding

    Yongxin Guo, Jingyu Liu, Mingda Li, Xiaoying Tang, Xi Chen, and Bo Zhao. Vtg-llm: Integrating timestamp knowledge into video llms for enhanced video temporal grounding. arXiv preprint arXiv:2405.13382, 2024a. Yongxin Guo, Jingyu Liu, Mingda Li, Xiaoying Tang, Qingbin Liu, and Xi Chen. Trace: Temporal grounding video llm via causal event modeling, 2024b. U...

    arXiv

  2. [9]

    Vtimellm: Empower llm to grasp video moments

    Bin Huang, Xin Wang, Hong Chen, Zihan Song, and Wenwu Zhu. Vtimellm: Empower llm to grasp video moments. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 14271–14280, 2024a. De-An Huang, Shijia Liao, Subhashree Radhakrishnan, Hongxu Yin, Pavlo Molchanov, Zhiding Yu, and Jan Kautz. Lita: Language instructed tempor...

    arXiv

  3. [10]

    Tvqa+: Spatio-temporal grounding for video question answering

    Jie Lei, Licheng Yu, Tamara L Berg, and Mohit Bansal. Tvqa+: Spatio-temporal grounding for video question answering. arXiv preprint arXiv:1904.11574,

    arXiv 1904

  4. [13]

    URL https://arxiv.org/abs/2306. 05424. WonJun Moon, Sangeek Hyun, SuBeen Lee, and Jae-Pil Heo. Correlation-guided query-dependency calibration in video representation learning for temporal grounding. arXiv preprint arXiv:2311.08835, 2023a. WonJun Moon, Sangeek Hyun, SangUk Park, Dongchan Park, and Jae-Pil Heo. Query-dependent video representation for mome...

    arXiv

  5. [14]

    Mengxue Qu, Xiaodong Chen, Wu Liu, Alicia Li, and Yao Zhao

    URL https://arxiv.org/abs/2402.11435. Mengxue Qu, Xiaodong Chen, Wu Liu, Alicia Li, and Yao Zhao. Chatvtg: Video temporal grounding via chat with video dialogue large language models,

    arXiv

  6. [15]

    ChatVTG: Video Temporal Grounding via Chat with Video Dialogue Large Language Models

    URL https://arxiv.org/abs/2410.12813. 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. In International conference on machine learning, pages 8748–8763. PMLR,

    work page Pith review arXiv

  7. [16]

    Yolov3: An incremental improvement

    Joseph Redmon. Yolov3: An incremental improvement. arXiv preprint arXiv:1804.02767,

    arXiv

  8. [17]

    Llama: Open and efficient foundation language models

    Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, et al. Llama: Open and efficient foundation language models. arXiv preprint arXiv:2302.13971,

    arXiv

Show all 25 references
  1. [20]

    Star: A benchmark for situated reasoning in real-world videos

    Bo Wu, Shoubin Yu, Zhenfang Chen, Joshua B Tenenbaum, and Chuang Gan. Star: A benchmark for situated reasoning in real-world videos. arXiv preprint arXiv:2405.09711,

    arXiv

  2. [21]

    When, where, and what? a new dataset for anomaly detection in driving videos

    15 Yu Yao, Xizi Wang, Mingze Xu, Zelin Pu, Ella Atkins, and David Crandall. When, where, and what? a new dataset for anomaly detection in driving videos. arXiv preprint arXiv:2004.03044,

    2004 arXiv

  3. [23]

    Timesuite: Improving mllms for long video understanding via grounded tuning, 2024a

    Xiangyu Zeng, Kunchang Li, Chenting Wang, Xinhao Li, Tianxiang Jiang, Ziang Yan, Songze Li, Yansong Shi, Zhengrong Yue, Yi Wang, Yali Wang, Yu Qiao, and Limin Wang. Timesuite: Improving mllms for long video understanding via grounded tuning, 2024a. URL https://arxiv.org/abs/24...

    arXiv

  4. [24]

    Video-llama: An instruction-tuned audio-visual language model for video understanding

    Hang Zhang, Xin Li, and Lidong Bing. Video-llama: An instruction-tuned audio-visual language model for video understanding. arXiv preprint arXiv:2306.02858, 2023b. Hao Zhang, Aixin Sun, Wei Jing, and Joey Tianyi Zhou. Span-based localizing network for natural language video lo...

    2004 arXiv

  5. [25]

    Deformable detr: Deformable transformers for end-to-end object detection

    Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, and Jifeng Dai. Deformable detr: Deformable transformers for end-to-end object detection. arXiv preprint arXiv:2010.04159,

    2010 arXiv

  6. [2005]

    Yolov4: Optimal speed and accuracy of object detection

    Alexey Bochkovskiy, Chien-Yao Wang, and Hong-Yuan Mark Liao. Yolov4: Optimal speed and accuracy of object detection. arXiv preprint arXiv:2004.10934,

    2004 arXiv

  7. [2010]

    Soda: Story oriented dense video captioning evaluation framework

    Soichiro Fujita, Tsutomu Hirao, Hidetaka Kamigaito, Manabu Okumura, and Masaaki Nagata. Soda: Story oriented dense video captioning evaluation framework. In Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part VI 16, pages 517...

    2020

  8. [2015]

    Grounded-videollm: Sharpening fine-grained temporal grounding in video large language models, 2024a

    Haibo Wang, Zhiyang Xu, Yu Cheng, Shizhe Diao, Yufan Zhou, Yixin Cao, Qifan Wang, Weifeng Ge, and Lifu Huang. Grounded-videollm: Sharpening fine-grained temporal grounding in video large language models, 2024a. URL https://arxiv.org/abs/2410.03290. Teng Wang, Ruimao Zhang, Zhi...

    arXiv

  9. [2016]

    Temporally grounding natural sentence in video

    Jingyuan Chen, Xinpeng Chen, Lin Ma, Zequn Jie, and Tat-Seng Chua. Temporally grounding natural sentence in video. In Proceedings of the 2018 conference on empirical methods in natural language processing, pages 162–171,

    2018

  10. [2017]

    Excl: Extractive clip localization using natural language descriptions

    Soham Ghosh, Anuva Agarwal, Zarana Parekh, and Alexander Hauptmann. Excl: Extractive clip localization using natural language descriptions. arXiv preprint arXiv:1904.02755,

    1904 arXiv

  11. [2018]

    Sharegpt4video: Improving video understanding and generation with better captions

    Lin Chen, Xilin Wei, Jinsong Li, Xiaoyi Dong, Pan Zhang, Yuhang Zang, Zehui Chen, Haodong Duan, Bin Lin, Zhenyu Tang, et al. Sharegpt4video: Improving video understanding and generation with better captions. arXiv preprint arXiv:2406.04325,

    arXiv

  12. [2019]

    Detecting moments and highlights in videos via natural language queries

    Jie Lei, Tamara L Berg, and Mohit Bansal. Detecting moments and highlights in videos via natural language queries. Advances in Neural Information Processing Systems, 34:11846–11858, 2021a. Jie Lei, Tamara L Berg, and Mohit Bansal. Detecting moments and highlights in videos via...

    arXiv

  13. [2020]

    Language models are few-shot learners

    Tom B Brown. Language models are few-shot learners. arXiv preprint arXiv:2005.14165,

    2005 arXiv

  14. [2021]

    Loconet: Long-short context network for active speaker detection

    Xizi Wang, Feng Cheng, and Gedas Bertasius. Loconet: Long-short context network for active speaker detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 18462–18472, 2024b. Yi Wang, Kunchang Li, Xinhao Li, Jiashuo Yu, Yinan He, ...

    arXiv

  15. [2022]

    Llava-next: Improved reasoning, ocr, and world knowledge, January 2024a

    Haotian Liu, Chunyuan Li, Yuheng Li, Bo Li, Yuanhan Zhang, Sheng Shen, and Yong Jae Lee. Llava-next: Improved reasoning, ocr, and world knowledge, January 2024a. URL https://llava-vl.github.io/ blog/2024-01-30-llava-next/ . Yuanxin Liu, Shicheng Li, Yi Liu, Yuxiang Wang, Shuhu...

    2024

  16. [2023]

    Aohan Zeng, Xiao Liu, Zhengxiao Du, Zihan Wang, Hanyu Lai, Ming Ding, Zhuoyi Yang, Yifan Xu, Wendi Zheng, Xiao Xia, et al

    URL https://arxiv.org/abs/2303.16406. Aohan Zeng, Xiao Liu, Zhengxiao Du, Zihan Wang, Hanyu Lai, Ming Ding, Zhuoyi Yang, Yifan Xu, Wendi Zheng, Xiao Xia, et al. Glm-130b: An open bilingual pre-trained model. arXiv preprint arXiv:2210.02414,

    arXiv

  17. [2024]

    Look closer to ground better: Weakly-supervised temporal grounding of sentence in video

    Zhenfang Chen, Lin Ma, Wenhan Luo, Peng Tang, and Kwan-Yee K Wong. Look closer to ground better: Weakly-supervised temporal grounding of sentence in video. arXiv preprint arXiv:2001.09308,

    2001 arXiv

Pith tools

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