Pith. sign in

REVIEW 3 major objections 5 minor 38 references

DRUM: Learning Demonstration Retriever for Large MUlti-modal Models

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read DRUM fine-tunes a multimodal embedding retriever with a list-wise ranking loss based on the LVLM's own conditional log-likelihood, and reports improved in-context learning on seven benchmark tasks.

desk verdict Equation 6 is sign-reversed: the loss trains the retriever to favor worse demonstrations, so the claimed improvements cannot be caused by the described optimization; the SIT-IPDR idea is worth salvaging, but the paper needs a major fix. read the letter →

arxiv 2412.07619 v1 pith:SFB6OJKC submitted 2024-12-10 cs.CL

Ellen Yi-Ge , Jiechao Gao , Wei Han , Wei Zhu This is my paper · ORCID
classification cs.CL
keywords in-contextlearninglargevision-languagemodelsdemonstrationretrievallist-wiserankinglossmultimodalembeddingsCLIPfine-tuningvisualquestionansweringimagecaptioning
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper claims that for in-context learning in large vision-language models (LVLMs), the choice of demonstrations should be learned from the LVLM's own preferences rather than fixed or similarity-based heuristics. DRUM fine-tunes a standard image-text embedding model so that its retrieval rankings match the LVLM's conditional log-likelihood of the correct answer under each candidate demonstration, using a list-wise ranking loss with rank-distance weights. At inference it retrieves demonstrations by concatenating image, prompt, and a two-token draft-response embedding for each test sample. On seven datasets spanning visual question answering, image classification, and image captioning, the fine-tuned retriever is reported to outperform random, fixed, CLIP-only, and EPR baselines, and to transfer to API-based LVLMs that cannot provide the training feedback themselves.

What carries the argument

The central object is the list-wise ranking loss $L_r = \sum_{i \neq j} m(i,j) \log(1 + e^{\text{sim}(x_q, z_j) - \text{sim}(x_q, z_i)})$, with $m(i,j) = \max(0, 1/\sqrt{r(z_j)} - 1/\sqrt{r(z_i)})$, where $r(z)$ is the rank assigned by the LVLM's conditional log-likelihood and $\text{sim}$ is the cosine similarity from the fine-tuned CLIP embedding model. This loss converts the LVLM's preference ordering into dense gradient signal for the retriever. The companion mechanism is the SIT-IPDR retrieval strategy, which concatenates image, prompt, and draft-response embeddings so that both modalities and the predicted answer contribute to similarity; a two-token draft response keeps the added inference cost small.

What would settle it

Compare the ranking produced by DRUM for a fixed set of queries when the query embedding is built from the ground-truth response versus from the two-token draft response. If the rank correlation or the downstream in-context learning accuracy drops markedly when the draft response replaces the ground truth, the train/test gap is real and load-bearing; measuring the cosine distance between the two query embeddings on a held-out set would also indicate whether the assumption holds.

Watch

Extended reading notes

Core claim

On its own terms, the paper's discovery is that a retriever can be trained to predict which demonstrations an LVLM will find useful, and that this prediction transfers across tasks and across LVLMs. Given a query sample, candidate demonstrations are ranked by the LVLM's conditional log-likelihood of the ground-truth response; the embedding model is trained with a list-wise ranking loss that pulls up the similarity of demonstrations the LVLM ranked higher and pushes down the rest, with per-pair weights proportional to the difference of inverse square roots of their ranks. The same loss works without task-specific modifications for visual question answering, image classification, and image captioning. The paper further claims that retrieval with concatenated image, prompt, and draft-response embeddings outperforms image-only or prompt-only variants, that iterative re-mining of candidates improves training, and that a retriever fine-tuned on feedback from an open LVLM improves the in-context performance of GPT-4o and Claude 3 Opus.

Load-bearing premise

At test time the query embedding uses a two-token draft response generated by the LVLM, but during training the query embedding includes the ground-truth response; the paper assumes these two representations place the query close enough in the embedding space that the fine-tuned rankings remain correct, and it provides no ablation or analysis of this mismatch.

Editorial extensions

If this is right

  • Demonstration quality can outweigh demonstration count: the paper reports DRUM with one or two demonstrations outperforming EPR with four on generation-heavy tasks.
  • The same list-wise ranking loss applies across task types without task-specific objectives, so new vision-language tasks need only a supporting set and a likelihood-capable LVLM to train a retriever.
  • Retrievers fine-tuned on one LVLM's feedback can improve other LVLMs, including API-only models that cannot provide likelihood feedback during training.
  • Combining image, prompt, and draft-response embeddings is the strongest retrieval configuration in the paper's ablations, suggesting text-only or image-only retrieval leaves useful signal unused.
  • Iterative demonstration candidate mining contributes to the gains, since removing it in the ablation lowers performance.

Reading between the lines

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

  • The unvalidated train/test gap around the draft response suggests a concrete extension: train with a small amount of response noise or a consistency regularizer so the retriever is robust to draft-response quality.
  • The same list-wise feedback loop could be applied to text-only LLM in-context learning or multimodal retrieval-augmented generation, wherever a frozen teacher can score candidate context blocks by likelihood.
  • The cross-LVLM transfer result implies demonstration preferences are partly shared across models; measuring rank correlation between different LVLMs on identical candidate sets would test how far that sharing extends.
  • Because the retriever is tuned to one LVLM's likelihood, updating the target LVLM may require re-mining candidates; the paper does not study how quickly DRUM's advantage decays across model generations.
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 / 5 minor

Summary. The paper proposes DRUM, a framework for fine-tuning a CLIP-based visual-language embedding model to retrieve in-context demonstrations for large vision-language models (LVLMs). The method defines retrieval strategies over images, prompts, and draft responses; uses the LVLM's conditional log-likelihood to rank candidate demonstrations; trains the retriever with a list-wise ranking loss; and iteratively mines demonstration candidates. Experiments are reported on seven benchmark datasets spanning VQA, image classification, and image captioning, using Open-Flamingo 9B as the inference LVLM, with claims of consistent improvement over baselines and transferability to commercial LVLMs.

Significance. If the described training objective were correct, DRUM would offer a general, task-agnostic way to adapt an embedding model to an LVLM's in-context learning preferences, which is a useful and timely contribution. The paper also addresses an important gap: most prior demonstration-retrieval work targets text-only LLMs, whereas this work targets multimodal LVLMs. However, the central training loss in Eq. (6) is, as written, anti-correlated with the LVLM feedback that it is supposed to inject, and the paper explicitly describes that anti-correlated behavior. The reported experimental gains therefore cannot be attributed to the described method. The paper also lacks a defined baseline, has a mismatched caption/table entry, and leaves a significant train/test representation gap unexamined. The contribution is not acceptable in its current form.

major comments (3)
  1. [Section 3.2.2, Eq. (6) and Eq. (7)] The list-wise ranking loss is sign-reversed relative to its stated purpose. For any pair with r(zj) < r(zi), meaning demonstration zj is ranked better by the LVLM, the coefficient m(i,j) = max(0, 1/sqrt(r(zj)) - 1/sqrt(r(zi))) is positive, and the term log(1 + exp(sim(xq,zj) - sim(xq,zi))) is minimized when sim(xq,zi) > sim(xq,zj). Thus the loss trains the retriever to assign higher similarity to the worse demonstration. The accompanying text confirms this: "Lr will draw sim(xq, zi) up and optimize the retriever towards sim(xq, zi) > sim(xq, zj)". This directly contradicts the stated goal of retrieving more helpful demonstrations and the validation metric S(E) in Eq. (10), which rewards agreement with the LVLM ranking. Since all ablation variants retain the same sign convention, the reported improvements cannot be causally explained by the described optimization. The authors must either correct the sign in Eq. (6) (and the surrounding text) or, if the implementation used a different objective, clearly restate the actual training objective and rerun the analysis.
  2. [Sections 3.1 and 3.2.1] There is an unexamined train/test gap in the query representation. During training, the querying example xq includes the ground-truth response responseq, and the retriever's similarity sim(xq, zj) is evidently based on the full triplet (imageq, promptq, responseq) under the SIT-IPDR strategy. At inference, the SIT-IPDR strategy uses a draft response of at most two tokens generated by the LVLM, which can be incomplete or incorrect. The paper provides no analysis, bound, or ablation showing that the learned retriever remains effective when the ground-truth response is replaced by this short draft. This gap is load-bearing because the retriever's ranking signal during training may rely heavily on the full response text, which is unavailable at deployment time.
  3. [Section 4.4, Table 1, and Section 5] The experimental presentation is internally inconsistent. Section 4.4 lists baselines (a) through (g) and does not define any method called "Dr-VL", yet Table 1 reports a "Dr-VL" row. The table caption states that results for "Random, Fixed, EPR, UDR and DRUM" are averaged over five runs, but the table contains no "UDR" row. The conclusion (Section 5) also says "UDR significantly outperforms the baseline demonstration retrieval methods" instead of DRUM. These inconsistencies make it impossible to determine which methods were actually compared and which of the reported numbers correspond to the authors' own prior UDR work. The authors must clarify the baseline set, correct the table and caption, and ensure the conclusion refers to the method proposed in this paper.
minor comments (5)
  1. [Section 3.2.2] The text refers to "w will be a high weight" when discussing m(i,j), but no variable w is defined; this should be m(i,j) throughout.
  2. [Section 3.2.2 and Section 3.2 (Eq. 9 vs Eq. 10)] Equation (9) and Equation (10) present the identical formula for S(E); the duplicate should be removed, and the validation metric should be defined once.
  3. [Section 4.3] The reference for Open-Flamingo is given as "Lin, 2004" (the ROUGE paper); the correct citation is Awadalla et al. (2023).
  4. [Throughout] There are several typos and inconsistent spellings: "Flicker30K" for Flickr30K, "lantency" for latency, "visiion-language" for vision-language, "concate" for concatenate, and "unifined" for unified. These should be corrected.
  5. [Section 5] The conclusion claims transferability across LVLMs from "4B to 175B", but the experiments use Open-Flamingo 9B, GPT-4o, and Claude 3 Opus; the paper does not report experiments with a 4B or 175B model, so this claim is unsupported.

Circularity Check

0 steps flagged · score 2.0 of 10

No load-bearing circularity: DRUM's retriever is empirically fitted to LVLM feedback and tested on held-out data; only a minor, non-load-bearing self-citation to UDR is present.

full rationale

The claimed derivation chain is empirical rather than definitional. Eq. 4 defines a demonstration's value by the LVLM's conditional log-likelihood of the ground-truth response; Eq. 6 ranks training pairs with a list-wise loss; Eq. 10 validates retriever consistency on a dev split; Table 1 then reports held-out LVLM accuracy on Dtest. No fitted parameter or objective is defined in terms of the reported test numbers, and Dclip_train, Dclip_dev, and Dtest are disjoint, so the reported improvement is not forced by construction. Using the same Open-Flamingo model for feedback and evaluation is a standard adaptation loop, not circularity, because the test queries are unseen and the reported metric (ICL accuracy) is not the training objective. The iterative mining in Sec. 3.3 is credited to Li et al. (2023c), a prior work sharing an author (Wei Zhu); however, this self-citation supplies only the candidate-update component (Eq. 8), is not used to justify the central effectiveness claim, and is ablated in DRUM-4, so it is not load-bearing. The apparent sign reversal in Eq. 6 (where, with r ascending so smaller r is better, the loss is minimized by making sim(xq, zi) > sim(xq, zj) for better zj) is a serious correctness/consistency issue if the equations match the implementation, but it is not a circularity pattern under the defined categories and therefore does not change this verdict.

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

The paper introduces no new physical or mathematical entities. The central claim relies on the LVLM's log-likelihood as a training signal, on the validity of concatenated CLIP embeddings, and on the assumption that a two-token draft response at inference behaves like the ground-truth response used in training. The hyperparameters listed are manually selected and not swept.

free parameters (4)
  • learning_rate = 1e-5
    Chosen by hand for fine-tuning the CLIP embedding model; no sensitivity analysis.
  • n_train_candidates = 32
    Number of retrieved candidates per query in training, set to 32; this affects the granularity of the ranking signal.
  • n_demonstrations = 4
    Number of demonstrations used at inference, set to 4 by default; results vary with this number.
  • training_iterations_and_steps = N1=50, N2=100
    Number of outer iterations and inner training steps; with early stopping patience 10, but chosen by hand.
assumptions (4)
  • domain assumption The LVLM's conditional log-likelihood of the ground-truth response provides a valid reward signal for ranking demonstration quality.
    Used in Eq 4 to define s(z_j) and produce training labels; no evidence that log-likelihood correlates with downstream accuracy.
  • domain assumption Concatenated CLIP image and text embeddings form a retrieval space in which cosine similarity is a meaningful measure of demonstration usefulness.
    Underlies the SIT-IPDR strategy and the fine-tuning objective; no analysis of how well CLIP geometry matches LVLM preferences.
  • domain assumption The two-token draft response at inference is sufficiently similar in embedding space to the training-time ground-truth response for the retriever to generalize.
    Train/test gap introduced by ST-PDR/SIT-IPDR; not validated.
  • domain assumption Iterative candidate mining with the same LVLM as scorer converges to a better training distribution.
    Adapted from Li et al. 2023c; no convergence analysis.

how reviews work

0 comments
Cite this review

Pith. "Pith review of DRUM: Learning Demonstration Retriever for Large MUlti-modal Models." pith.science (2026). https://pith.science/paper/SFB6OJKC

@misc{pith2026241207619,
  author       = {Pith},
  title        = {Pith review of: DRUM: Learning Demonstration Retriever for Large MUlti-modal Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SFB6OJKC}},
  note         = {Machine review of arXiv:2412.07619}
}
read the original abstract

Recently, large language models (LLMs) have demonstrated impressive capabilities in dealing with new tasks with the help of in-context learning (ICL). In the study of Large Vision-Language Models (LVLMs), when implementing ICL, researchers usually adopts the naive strategies like fixed demonstrations across different samples, or selecting demonstrations directly via a visual-language embedding model. These methods does not guarantee the configured demonstrations fit the need of the LVLMs. To address this issue, we now propose a novel framework, \underline{d}emonstration \underline{r}etriever for large m\underline{u}lti-modal \underline{m}odel (DRUM), which fine-tunes the visual-language embedding model to better meet the LVLM's needs. First, we discuss the retrieval strategies for a visual-language task, assuming an embedding model is given. And we propose to concate the image and text embeddings to enhance the retrieval performance. Second, we propose to re-rank the demonstrations retrieved by the embedding model via the LVLM's feedbacks, and calculate a list-wise ranking loss for training the embedding model. Third, we propose an iterative demonstration mining strategy to improve the training of the embedding model. Through extensive experiments on 3 types of visual-language tasks, 7 benchmark datasets, our DRUM framework is proven to be effective in boosting the LVLM's in-context learning performance via retrieving more proper demonstrations.

Figures

Figures reproduced from arXiv: 2412.07619 by the authors.

Figure 1
Figure 1. The schematic representation of our DRUM framework. Circles, rectangles, and triangles respectively [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. The effects of the number of demonstrations on DRUM, EPR, and CLIP. [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 5 canonical work pages

  1. [1]

    URL: " 'urlintro :=

    ENTRY address author booktitle chapter edition editor howpublished institution journal key month note number organization pages publisher school series title type volume year eprint doi pubmed url lastchecked label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block STRINGS urlintro eprinturl eprintpr...

  2. [2]

    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 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Harsh Agrawal, Karan Desai, Yufei Wang, Xinlei Chen, Rishabh Jain, Mark Johnson, Dhruv Batra, Devi Parikh, Stefan Lee, and Peter Anderson. 2019. Nocaps: Novel object captioning at scale. In Proceedings of the IEEE/CVF international conference on computer vision, pages 8948--8957

    2019

  4. [4]

    Jean-Baptiste Alayrac, Jeff Donahue, Pauline Luc, Antoine Miech, Iain Barr, Yana Hasson, Karel Lenc, Arthur Mensch, Katherine Millican, Malcolm Reynolds, et al. 2022. Flamingo: a visual language model for few-shot learning. Advances in neural information processing systems, 35:23716--23736

    2022

  5. [5]

    Anas Awadalla, Irena Gao, Josh Gardner, Jack Hessel, Yusuf Hanafy, Wanrong Zhu, Kalyani Marathe, Yonatan Bitton, Samir Gadre, Shiori Sagawa, et al. 2023. Openflamingo: An open-source framework for training large autoregressive vision-language models. arXiv preprint arXiv:2308.01390

    arXiv 2023

  6. [6]

    Jinze Bai, Shuai Bai, Shusheng Yang, Shijie Wang, Sinan Tan, Peng Wang, Junyang Lin, Chang Zhou, and Jingren Zhou. 2023. Qwen-vl: A frontier large vision-language model with versatile abilities. arXiv preprint arXiv:2308.12966

    arXiv 2023

  7. [7]

    Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. 2020. Language models are few-shot learners. Advances in neural information processing systems, 33:1877--1901

    2020

  8. [8]

    Mathilde Caron, Hugo Touvron, Ishan Misra, Herv \'e J \'e gou, Julien Mairal, Piotr Bojanowski, and Armand Joulin. 2021. Emerging properties in self-supervised vision transformers. In Proceedings of the IEEE/CVF international conference on computer vision, pages 9650--9660

    2021

Show all 38 references
  1. [9]

    Danqi Chen, Adam Fisch, Jason Weston, and Antoine Bordes. 2017. Reading wikipedia to answer open-domain questions. arXiv preprint arXiv:1704.00051

    2017 arXiv

  2. [10]

    Jianlyu Chen, Shitao Xiao, Peitian Zhang, Kun Luo, Defu Lian, and Zheng Liu. 2024. M3-embedding: Multi-linguality, multi-functionality, multi-granularity text embeddings through self-knowledge distillation. In Findings of the Association for Computational Linguistics ACL 2024,...

    2024

  3. [11]

    Matthijs Douze, Alexandr Guzhva, Chengqi Deng, Jeff Johnson, Gergely Szilvasy, Pierre-Emmanuel Mazar \'e , Maria Lomeli, Lucas Hosseini, and Herv \'e J \'e gou. 2024. The faiss library. arXiv preprint arXiv:2401.08281

    2024 arXiv

  4. [12]

    Yifan Du, Zikang Liu, Junyi Li, and Wayne Xin Zhao. 2022. A survey of vision-language pre-trained models. arXiv preprint arXiv:2202.10936

    2022 arXiv

  5. [13]

    Yash Goyal, Tejas Khot, Douglas Summers-Stay, Dhruv Batra, and Devi Parikh. 2017. Making the v in vqa matter: Elevating the role of image understanding in visual question answering. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 6904--6913

    2017

  6. [14]

    Danna Gurari, Qing Li, Abigale J Stangl, Anhong Guo, Chi Lin, Kristen Grauman, Jiebo Luo, and Jeffrey P Bigham. 2018. Vizwiz grand challenge: Answering visual questions from blind people. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 3...

    2018

  7. [15]

    Neil Houlsby, Andrei Giurgiu, Stanislaw Jastrzebski, Bruna Morrone, Quentin De Laroussilhe, Andrea Gesmundo, Mona Attariyan, and Sylvain Gelly. 2019. Parameter-efficient transfer learning for nlp. In International Conference on Machine Learning, pages 2790--2799. PMLR

    2019

  8. [16]

    Douwe Kiela, Hamed Firooz, Aravind Mohan, Vedanuj Goswami, Amanpreet Singh, Pratik Ringshia, and Davide Testuggine. 2020. The hateful memes challenge: Detecting hate speech in multimodal memes. Advances in neural information processing systems, 33:2611--2624

    2020

  9. [17]

    Junnan Li, Dongxu Li, Silvio Savarese, and Steven Hoi. 2023 a . Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models. In International conference on machine learning, pages 19730--19742. PMLR

    2023

  10. [18]

    Li Li, Jiawei Peng, Huiyi Chen, Chongyang Gao, and Xu Yang. 2024. How to configure good in-context sequence for visual question answering. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 26710--26720

    2024

  11. [19]

    Rui Li, Guoyin Wang, and Jiwei Li. 2023 b . Are human-generated demonstrations necessary for in-context learning? arXiv preprint arXiv:2309.14681

    2023 arXiv

  12. [20]

    Xiaonan Li, Kai Lv, Hang Yan, Tianya Lin, Wei Zhu, Yuan Ni, Guo Tong Xie, Xiaoling Wang, and Xipeng Qiu. 2023 c . https://api.semanticscholar.org/CorpusID:258557751 Unified demonstration retriever for in-context learning . ArXiv, abs/2305.04320

    2023 arXiv

  13. [21]

    Chin-Yew Lin. 2004. Rouge: A package for automatic evaluation of summaries. In Text summarization branches out, pages 74--81

    2004

  14. [22]

    Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. 2024. Visual instruction tuning. Advances in neural information processing systems, 36

    2024

  15. [23]

    Ilya Loshchilov and Frank Hutter. 2019. Decoupled weight decay regularization. In ICLR

    2019

  16. [24]

    Xinxi Lyu, Sewon Min, Iz Beltagy, Luke Zettlemoyer, and Hannaneh Hajishirzi. 2022. Z-icl: Zero-shot in-context learning with pseudo-demonstrations. arXiv preprint arXiv:2212.09865

    2022 arXiv

  17. [25]

    Kenneth Marino, Mohammad Rastegari, Ali Farhadi, and Roozbeh Mottaghi. 2019. Ok-vqa: A visual question answering benchmark requiring external knowledge. In Proceedings of the IEEE/cvf conference on computer vision and pattern recognition, pages 3195--3204

    2019

  18. [26]

    Marius Mosbach, Tiago Pimentel, Shauli Ravfogel, Dietrich Klakow, and Yanai Elazar. 2023. Few-shot fine-tuning vs. in-context learning: A fair comparison and evaluation. arXiv preprint arXiv:2305.16938

    2023 arXiv

  19. [27]

    Maria-Elena Nilsback and Andrew Zisserman. 2008. Automated flower classification over a large number of classes. In 2008 Sixth Indian conference on computer vision, graphics & image processing, pages 722--729. IEEE

    2008

  20. [28]

    Jane Pan. 2023. What in-context learning “learns” in-context: Disentangling task recognition and task learning. Master's thesis, Princeton University

    2023

  21. [29]

    Bryan A Plummer, Liwei Wang, Chris M Cervantes, Juan C Caicedo, Julia Hockenmaier, and Svetlana Lazebnik. 2015. Flickr30k entities: Collecting region-to-phrase correspondences for richer image-to-sentence models. In Proceedings of the IEEE international conference on computer ...

    2015

  22. [30]

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

    2021

  23. [31]

    Ohad Rubin, Jonathan Herzig, and Jonathan Berant. 2021. Learning to retrieve prompts for in-context learning. arXiv preprint arXiv:2112.08633

    2021 arXiv

  24. [32]

    Jason Wei, Yi Tay, Rishi Bommasani, Colin Raffel, Barret Zoph, Sebastian Borgeaud, Dani Yogatama, Maarten Bosma, Denny Zhou, Donald Metzler, et al. 2022. Emergent abilities of large language models. arXiv preprint arXiv:2206.07682

    2022 arXiv

  25. [33]

    Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pierric Cistac, Tim Rault, Rémi Louf, Morgan Funtowicz, Joe Davison, Sam Shleifer, Patrick von Platen, Clara Ma, Yacine Jernite, Julien Plu, Canwen Xu, Teven Le Scao, Sylvain Gugger, Mari...

    2020

  26. [34]

    Xu Yang, Yongliang Wu, Mingzhuo Yang, Haokun Chen, and Xin Geng. 2024. Exploring diverse in-context configurations for image captioning. Advances in Neural Information Processing Systems, 36

    2024

  27. [35]

    Deyao Zhu, Jun Chen, Xiaoqian Shen, Xiang Li, and Mohamed Elhoseiny. 2023. Minigpt-4: Enhancing vision-language understanding with advanced large language models. arXiv preprint arXiv:2304.10592

    2023 arXiv

  28. [36]

    @esa (Ref

    \@ifxundefined[1] #1\@undefined \@firstoftwo \@secondoftwo \@ifnum[1] #1 \@firstoftwo \@secondoftwo \@ifx[1] #1 \@firstoftwo \@secondoftwo [2] @ #1 \@temptokena #2 #1 @ \@temptokena \@ifclassloaded agu2001 natbib The agu2001 class already includes natbib coding, so you should ...

  29. [37]

    \@lbibitem[] @bibitem@first@sw\@secondoftwo \@lbibitem[#1]#2 \@extra@b@citeb \@ifundefined br@#2\@extra@b@citeb \@namedef br@#2 \@nameuse br@#2\@extra@b@citeb \@ifundefined b@#2\@extra@b@citeb @num @parse #2 @tmp #1 NAT@b@open@#2 NAT@b@shut@#2 \@ifnum @merge>\@ne @bibitem@firs...

  30. [38]

    @open @close @open @close and [1] URL: #1 \@ifundefined chapter * \@mkboth \@ifxundefined @sectionbib * \@mkboth * \@mkboth\@gobbletwo \@ifclassloaded amsart * \@ifclassloaded amsbook * \@ifxundefined @heading @heading NAT@ctr thebibliography [1] @ \@biblabel @NAT@ctr \@bibset...

Pith tools

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