Align, Generate, Learn: A Novel Closed-Loop Framework for Cross-Lingual In-Context Learning

Mateo Alejandro Rojas; Rafael Carranza

REVIEW 6 major objections 5 minor 16 references

Align, Generate, Learn: A Novel Closed-Loop Framework for Cross-Lingual In-Context Learning

T0 review · 6 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 A language model that generates its own in-context examples outperforms retrieval-based baselines on multilingual tasks.

desk verdict A self-supervised XICL framework whose headline claim is undercut by an undefined sampling distribution over example sets; the equations do not cohere. read the letter →

arxiv 2412.08955 v1 pith:HET6KPFB submitted 2024-12-12 cs.CL

Mateo Alejandro Rojas , Rafael Carranza This is my paper · ORCID

classification cs.CL

keywords cross-lingualin-contextlearninglargelanguagemodelslow-resourcelanguagesretrieval-generationalignmentsemanticcoherencereinforcementself-supervisedtrainingmultilingualbenchmarks

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper argues that cross-lingual in-context learning can be driven entirely by the LLM's own generative abilities, without external retrievers or task-specific fine-tuning. It proposes a closed-loop training scheme in which the model generates candidate example sets for a query, aligns retrieval and generation through a KL-divergence objective, enforces cross-lingual semantic coherence in embedding space, and refines example selection with reinforcement learning. The reported experiments put this framework ahead of the strongest baseline, XICL-RL, on average accuracy (76.1% vs. 69.5%), with larger margins in low-resource languages, and the ablation study attributes the gain to both training objectives. If correct, the result would make multilingual few-shot learning more scalable by removing the retrieval component from the pipeline.

What carries the argument

The load-bearing object is the closed-loop training objective $\mathcal{L} = \mathcal{L}_{\text{align}} + \lambda \mathcal{L}_{\text{coherence}} + \gamma \mathcal{L}_{\text{RL}}$. $\mathcal{L}_{\text{align}}$ pushes the model's output distribution with selected examples toward its output distribution without them, so the chosen examples carry information rather than noise; $\mathcal{L}_{\text{coherence}}$ pulls the encoder representations of input and output together across languages; $\mathcal{L}_{\text{RL}}$ rewards example sets that improve task accuracy and semantic diversity. The design calls for the model to generate synthetic example pairs $C$ itself, evaluate them with the reward, and update $\theta$, turning example selection into a trainable behavior.

What would settle it

Run the same three-term training loss on a fixed backbone while replacing the model-generated example sets with random subsets of the training corpus. If accuracy stays near the reported 76.1%, the alignment and coherence losses alone explain the gain; if it drops toward the 69.5% baseline, the internal example-generation mechanism is doing the work. Either outcome would pin down what the closed-loop claim actually contributes.

Watch

Extended reading notes

Core claim

The central claim is that an LLM can be trained to internally select and exploit task-relevant examples across languages, making an external retrieval step unnecessary. Formally, the paper defines a generative model $p_\theta(y \mid x, C)$ for output $y$ given input $x$ and example set $C$, and optimizes three objectives: a retrieval-generation alignment loss $\mathcal{L}_{\text{align}} = \mathrm{KL}(p_\theta(y \mid x, C) \| p_\theta(y \mid x))$, a semantic coherence loss on encoder embeddings $\|h_x - h_y\|^2$, and a reinforcement-learning objective that maximizes a reward $R(x, C) = \alpha A(x, C) + \beta D(C)$ combining task accuracy and semantic diversity. The reported outcome is a consistent accuracy improvement over random sampling, SBERT retrieval, Glot500 RET, and the XICL-RL baseline, with the largest gains on low-resource and typologically distant languages; human evaluation and unseen-task tests are offered as supporting evidence.

Load-bearing premise

The framework's main training loop requires the model to sample sets of example inputs and outputs from its own distribution, but the paper never explains how those example sets are produced, so the mechanism that supposedly carries the gain is left undefined.

Editorial extensions

If this is right

External retrievers become optional: a single trained LLM could select its own in-context examples at inference time.
The framework should transfer to unseen tasks such as summarization and named entity recognition without fine-tuning, as the paper's generalization experiments claim.
Low-resource and typologically diverse languages stand to gain the most, since the reported margins over baselines are largest there.
Both training objectives are load-bearing: dropping either one lowers overall accuracy by roughly three to four points in the ablation.
Human-rated relevance, fluency, and semantic correctness all improve over the strongest baseline, supporting practical usability beyond accuracy.

Reading between the lines

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

An implicit testable extension: if closed-loop example selection is the true source of the gain, the same training scheme should also improve monolingual in-context learning, since nothing in the objective is cross-lingual-specific except the evaluation.
Because the paper never specifies how the example set $C$ is sampled, a fair replication would need to fix that procedure; the reported gains may depend on a particular generation heuristic rather than on the loss formulas alone.
A direct comparison with an oracle retriever using gold examples would show whether the internal selector matches, approaches, or falls short of external selection, clarifying what the closed-loop claim actually buys.
The diversity term in the reward predicts that example sets with broader linguistic coverage improve accuracy; this is measurable by varying $\beta$ and watching low-resource performance.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Quick take: the reader's REJECT is right, and the stress-test pinpoints the load-bearing flaw. Eq. (4) samples example sets C from pθ, but pθ is only defined for token sequences y given x and C. There is no generative model over C anywhere in the paper, no marginalization, no sampling procedure. That means the RL objective, the alignment loss, and the whole reported accuracy gain are not well-defined. This is not a stylistic gap; the method cannot be instantiated as written.

To give credit where it's due, the high-level problem is real: XICL typically relies on external retrievers or task-specific fine-tuning, and a retriever-free, self-supervised way to let the LLM internally select examples would be useful. The paper is clearly written at the level of prose, and the related work covers the relevant threads. The ablation design is sensible in spirit.

But the gaps are everywhere. Eq. (3) calls it an encoder when the model is a decoder-only LLM — a direct contradiction. Eq. (5) uses task accuracy A(x,C), which requires labels, yet the pipeline is called self-supervised. The experiments report numbers without naming datasets, models, or hyperparameters, and without error bars, so reproduction is impossible. And Table 6, labeled low-resource, shows English outputs. It reads like the experimental section was assembled to a template.

Who is this for? Maybe a reader wanting a sketch of what a closed-loop XICL could look like, but not anyone wanting a working method. The citations are fine. The math is not. A serious referee could help the authors, but the paper in its current form has too many undefined core objects to justify sending out; it should go back for a complete rewrite and real experiments. If a revised version defines the sampling distribution, replaces the encoder with a decoder-compatible representation, reconciles self-supervision with task accuracy, and reports a concrete experimental setup, I'd revisit.

Referee Report

6 major / 5 minor

Summary. The manuscript proposes a closed-loop, self-supervised framework for cross-lingual in-context learning (XICL). The high-level idea is to train a large language model to generate its own in-context example sets, using a retrieval-generation alignment loss, a semantic coherence loss, and a reinforcement-learning objective that rewards task accuracy and semantic diversity. The authors report state-of-the-art results on a multilingual benchmark, an ablation study, human evaluation, and generalization to unseen tasks. However, the formal description of the training objective is underspecified and internally inconsistent: the model distribution is defined only over output sequences, the coherence loss assumes an encoder that is never introduced, and the reward requires labels in a pipeline claimed to be self-supervised. The experimental section also omits the dataset, base model, and hyperparameters, so the reported numbers cannot be verified.

Significance. If the proposed training pipeline were fully specified and validated, the idea of having an LLM generate and refine its own in-context examples could be a useful step toward removing external retrievers in cross-lingual settings. The paper identifies a real limitation and has a plausible high-level direction. However, as submitted, the significance cannot be assessed because the central objectives are not well-defined and the empirical support is incomplete. The reported accuracy gains, ablation trends, and human evaluation scores are all attributed to a procedure that is not reproducible from the text.

major comments (6)

[§3.3, Eq. (4)] Eq. (4) defines L_RL = -E_{x,C∼pθ}[R(x,C)], treating pθ as a distribution over example sets C. Yet Eq. (1) defines pθ only as a conditional distribution over output token sequences y given x and C: pθ(y | x, C). No generative process, marginalization, or sampling scheme for C is specified anywhere in the paper; in particular, the text does not state what prompt, decoding scheme, or temperature is used to produce C, nor how a distribution over sets of pairs would be normalized. Because the RL objective and the alignment objective in Eq. (2) both depend on this undefined distribution, the overall loss in Eq. (6) cannot be instantiated, and the experimental results in Tables 1–5 are not grounded in a well-defined training procedure.
[§3.2, Eq. (3)] Eq. (3) requires computing semantic embeddings hx and hy with 'the model's encoder,' but the model is presented in §3.1 as a generative sequence model pθ(y | x, C) with no encoder component. The paper never defines how hx and hy are obtained, what the encoder architecture is, or how its parameters relate to θ. In addition, because the loss directly minimizes the squared ℓ2 distance between hx and hy, the claimed 'cross-lingual consistency' is enforced by construction rather than discovered; a separate, non-training evaluation would be needed to support the paper's consistency claims.
[§3.3, Eq. (5)] The reward in Eq. (5), R(x,C)=αA(x,C)+βD(C), depends on task accuracy A(x,C), which by standard definition requires ground-truth labels for the query x. This conflicts with the paper's repeated claim that the framework is self-supervised (see Abstract, §1, and §3.2). The authors must explain how A(x,C) is evaluated without labeled data, or revise the self-supervision claim. The term D(C) ('semantic diversity') is also never formally defined.
[§4.1] The experimental section does not provide the dataset name, the base LLM, the model size, the number of languages and tasks, the train/validation splits, or the values of the hyperparameters α, β, λ, and γ. It also does not describe the optimization procedure (optimizer, learning rate, number of steps, compute). The sentence in §4.1 that all models use 'consistent hyperparameters' does not make the experiments reproducible, and it is impossible to tell whether the comparison in Tables 1–5 is controlled.
[§4.2, Table 1] The strongest baseline, XICL-RL, is used in Tables 1, 3, and 5, but it is neither defined nor cited anywhere in the paper; 'Glot500 RET' is likewise introduced without a reference or implementation description. Without a specification of these baselines, the claimed state-of-the-art improvement in Table 1 cannot be evaluated.
[Tables 1–3] Tables 1–3 report only point estimates with no standard deviations, confidence intervals, or significance tests, yet the abstract and §4.2 use the word 'significantly.' The claims of significance and of robustness across language families (Table 4) are not supported by the reported statistics.

minor comments (5)

[§3.2, Eq. (2)] The expectation E_{x,C} does not specify the distribution over x or C; even apart from the missing distribution over C, all expectations in the paper should be written with explicit probability measures.
[Tables 1 and 5] Table 1 is captioned as average accuracy, Table 5 reports F1, and the text mentions BLEU for translation tasks, but no translation experiments are shown; the metrics used in each table should be defined in the experimental setup.
[Table 4] Table 4 gives aggregate accuracy by language family without sample sizes or variance; the claim that the method is robust across families needs per-language or at least per-family breakdowns with counts.
[Table 6] Table 6 is described as a text classification example, but the generated outputs and ground truth are full sentences rather than class labels; the example should be reconciled with the task or replaced.
[References] Some cited references appear unrelated to the claims they support; for example, Bellström and Kilbrink (2008), a paper on problem-based learning in programming, is cited for the cross-lingual capabilities of large language models in the Introduction. The citation list should be checked carefully.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the paper's objectives directly define their targets, and no prediction reduces to an input or to a self-citation chain.

full rationale

I walked the derivation chain from Eq. (1) through Eq. (6) and the experimental section. The semantic coherence loss (Eq. 3) and the RL reward (Eq. 5) are training objectives, not predictions: reporting higher accuracy after directly optimizing task accuracy in the reward is standard supervised/RL evaluation, not a circular reduction. The retrieval-generation alignment loss (Eq. 2) is likewise an objective designed to make outputs insensitive to example-set variation; any claim that it 'ensures consistency' is a statement about the training goal, not an independently derived result. There are no load-bearing self-citations: the cited Zhou et al. works are by different authors and are used for background, not to justify the framework. No uniqueness theorem is imported, no ansatz is smuggled via citation, and no known result is merely renamed. The one substantive concern is formal, not circular: Eq. (4) writes C ~ p_theta even though p_theta is defined in Eq. (1) only as a distribution over output sequences y, so the sampling process for C is underspecified; this is a rigor/reproducibility gap, not a reduction of a claimed result to its own input. Consequently, the paper's central claims do not reduce by construction to their inputs.

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

The framework depends on four unspecified hyperparameters and on three assumptions about the model that are not established by the definition of pθ(y|x,C). No new physical or conceptual entities are introduced.

free parameters (4)

alpha
Weight for task accuracy in reward function R(x,C) in Eq. (5); no value or tuning procedure given.
beta
Weight for semantic diversity in Eq. (5); no value given.
lambda
Weight for coherence loss in Eq. (6); no value given.
gamma
Weight for RL loss in Eq. (6); no value given.

assumptions (4)

ad hoc to paper The model pθ can sample synthetic example-query pairs C, as in Eq. (4), although the model is defined as pθ(y | x, C).
Eq. (4) writes C ~ pθ, but the only distribution defined is over output tokens y; no generative process for C is described.
ad hoc to paper The model has an encoder that produces semantic embeddings hx and hy for input and output sequences.
Eq. (3) requires hidden embeddings, but the model is introduced as a generative sequence model without an encoder.
ad hoc to paper Task accuracy A(x,C) is available in a self-supervised setting without labeled data.
Section 3.3 calls the pipeline self-supervised yet defines reward based on task accuracy, which normally requires ground truth; the source of supervision is unspecified.
standard math KL divergence in Eq. (2) is finite and differentiable for the model distributions.
The loss uses KL(pθ(y|x,C) || pθ(y|x)), a standard divergence, but convergence and differentiability are not discussed.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Align, Generate, Learn: A Novel Closed-Loop Framework for Cross-Lingual In-Context Learning." pith.science (2026). https://pith.science/paper/HET6KPFB

@misc{pith2026241208955,
  author       = {Pith},
  title        = {Pith review of: Align, Generate, Learn: A Novel Closed-Loop Framework for Cross-Lingual In-Context Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HET6KPFB}},
  note         = {Machine review of arXiv:2412.08955}
}

read the original abstract

Cross-lingual in-context learning (XICL) has emerged as a transformative paradigm for leveraging large language models (LLMs) to tackle multilingual tasks, especially for low-resource languages. However, existing approaches often rely on external retrievers or task-specific fine-tuning, limiting their scalability and generalizability. In this paper, we propose a novel self-supervised framework that harnesses the generative capabilities of LLMs to internally select and utilize task-relevant examples. Our method introduces two key objectives: a retrieval-generation alignment loss to optimize the quality of selected examples and a semantic coherence loss to ensure cross-lingual consistency. Through extensive experiments on multilingual benchmarks, our approach achieves state-of-the-art performance, significantly outperforming existing baselines. Further analysis highlights its robustness across diverse language families and its ability to generalize to unseen tasks. Human evaluations confirm the superior fluency, relevance, and semantic correctness of outputs generated by our method. This work provides a scalable, effective, and generalizable solution for cross-lingual in-context learning.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

16 extracted references · 3 canonical work pages

[1]

Problem-based learnin g in a programming context-planning and executing a pilot survey on database access in a programming language

Peter Bellstr¨ om and Nina Kilbrink. Problem-based learnin g in a programming context-planning and executing a pilot survey on database access in a programming language. In George Angelos Pa- padopoulos, Wita Wojtkowski, W. Gregory Wojtkowski, Stani slaw Wrycza, and Joze Zupancic, ed- itors, Information Systems Development, Towards a Service Provis ion So...

work page 2008
[5]

URLhttps://doi.org/10.48550/arXiv.2303.12961

doi: 10.48550/ARXIV .2303.12961. URLhttps://doi.org/10.48550/arXiv.2303.12961. Humza Naveed, Asad Ullah Khan, Shi Qiu, Muhammad Saqib, Saee d Anwar, Muham- mad Usman, Nick Barnes, and Ajmal Mian. A comprehensive over view of large lan- guage models. CoRR, abs/2307.06435,

arXiv doi:10.48550/arxiv.2303.12961
[6]

URL https://doi.org/10.48550/arXiv.2307.06435

doi: 10.48550/ARXIV .2307.06435. URL https://doi.org/10.48550/arXiv.2307.06435. Zixuan Zhou, Xuefei Ning, Ke Hong, Tianyu Fu, Jiaming Xu, Shi yao Li, Y uming Lou, Luning Wang, Zhihang Y uan, Xiuhong Li, Shengen Y an, Guohao Dai, Xiao-Ping Zhang, Y uhan Dong, and Y u Wang. A survey on efﬁcient inference for large language models. CoRR, abs/2404.14294, 2024c. doi:

arXiv doi:10.48550/arxiv.2307.06435
[9]

URL https://doi.org/10.1145/3624724

doi: 10.1145/3624724. URL https://doi.org/10.1145/3624724. Y ucheng Zhou and Guodong Long. Improving cross-modal align ment for text-guided image inpainting. In Proceedings of the 17th Conference of the European Chapter o f the Association for Computational Linguistics, pages 3445–3456, 2023a. Y ucheng Zhou and Guodong Long. Multimodal event transformer f...

doi:10.1145/3624724
[10]

URLhttps://doi.org/10.48550/arXiv.2404.14294

48550/ARXIV .2404.14294. URLhttps://doi.org/10.48550/arXiv.2404.14294. Y ucheng Zhou, Zhi Rao, Jun Wan, and Jianbing Shen. Rethinkin g visual dependency in long-context reasoning for large vision-language models. arXiv preprint arXiv:2410.19732, 2024d. Yiran Zhao, Wenxuan Zhang, Guizhen Chen, Kenji Kawaguchi, a nd Lidong Bing. How do large lan- guage mode...

arXiv doi:10.48550/arxiv.2404.14294
[11]

Amanda Bertsch, Maor Ivgi, Uri Alon, Jonathan Berant, Matth ew R

URL https://openreview.net/forum?id=RdJVFCHjUMI. Amanda Bertsch, Maor Ivgi, Uri Alon, Jonathan Berant, Matth ew R. Gormley, and Graham Neubig. In-context learning with long-context mode ls: An in-depth explo- ration. CoRR, abs/2405.00200,

arXiv
[12]

URL https://doi.org/10.48550/arXiv.2405.00200

doi: 10.48550/ARXIV .2405.00200. URL https://doi.org/10.48550/arXiv.2405.00200. Yixing Jiang, Jeremy Irvin, Ji Hun Wang, Muhammad Ahmed Chau dhry, Jonathan H. Chen, and Andrew Y . Ng. Many-shot in-context learning in mul timodal foundation models. CoRR, abs/2405.09798,

arXiv doi:10.48550/arxiv.2405.00200
[13]

URL https://doi.org/10.48550/arXiv.2405.09798

doi: 10.48550/ARXIV .2405.09798. URL https://doi.org/10.48550/arXiv.2405.09798. Yingcong Li, Muhammed Emrullah Ildiz, Dimitris S. Papailio poulos, and Samet Oymak. Transformers as algorithms: Generalization and implicit m odel selection in in-context learning. CoRR, abs/2301.07067,

arXiv doi:10.48550/arxiv.2405.09798

Show all 16 references

[14]

URL https://doi.org/10.48550/arXiv.2301.07067

doi: 10.48550/ARXIV .2301.07067. URL https://doi.org/10.48550/arXiv.2301.07067. Tianjun Zhang, Aman Madaan, Luyu Gao, Steven Zheng, Swaroop Mishra, Yiming Y ang, Niket Tandon, and Uri Alon. In-context principle learning from mistakes. In F orty-ﬁrst International Conference on...

doi arXiv
[15]

Y ufeng Zhang, Fengzhuo Zhang, Zhuoran Y ang, and Zhaoran Wan g

URL https://openreview.net/forum?id=PAPY0cAB3C. Y ufeng Zhang, Fengzhuo Zhang, Zhuoran Y ang, and Zhaoran Wan g. What and how does in-context learning learn? bayesian model averaging, parameterization, and gener- alization. CoRR, abs/2305.19420,

arXiv
[16]

URL https://doi.org/10.48550/arXiv.2305.19420

doi: 10.48550/ARXIV .2305.19420. URL https://doi.org/10.48550/arXiv.2305.19420. 9

doi arXiv
[91]

Y ucheng Zhou, Xiang Li, Qianning Wang, and Jianbing Shen

URL https://doi.org/10.1007/b137171_91. Y ucheng Zhou, Xiang Li, Qianning Wang, and Jianbing Shen. Vi sual in-context learning for large vision-language models. In Findings of the Association for Computational Linguistics , ACL 2024, Bangkok, Thailand and virtual meeting, Augu...

doi
[2008]

doi: 10.1007/B137171\

doi
[2022]

NAACL-MAIN.201

doi: 10.18653/V1/2022. NAACL-MAIN.201. URL https://doi.org/10.18653/v1/2022.naacl-main.201. Sang Michael Xie, Aditi Raghunathan, Percy Liang, and Tengy u Ma. An explanation of in- context learning as implicit bayesian inference. In The Tenth International Conference on Learn- ...

2022 doi
[2023]

7 Y ucheng Zhou, Xiubo Geng, Tao Shen, Chongyang Tao, Guodong Long, Jian-Guang Lou, and Jianbing Shen

URL https://openreview.net/forum?id=nF70Sl-HUZ. 7 Y ucheng Zhou, Xiubo Geng, Tao Shen, Chongyang Tao, Guodong Long, Jian-Guang Lou, and Jianbing Shen. Thread of thought unraveling chaotic contexts. arXiv preprint arXiv:2311.08734, 2023a. Y ucheng Zhou, Tao Shen, Xiubo Geng, Ch...

2023 arXiv
[2024]

doi: 10.48550/ARXIV .2402. 18815. URL https://doi.org/10.48550/arXiv.2402.18815. Pei Zhou, Aman Madaan, Srividya Pranavi Potharaju, Aditya Gupta, Kevin R. McKee, Ari Holtzman, Jay Pujara, Xiang Ren, Swaroop Mishra, Aida Nematzadeh, Shyam U padhyay, and Manaal Faruqui. How far ...

doi arXiv

Pith tools

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

[1] [1]

Problem-based learnin g in a programming context-planning and executing a pilot survey on database access in a programming language

Peter Bellstr¨ om and Nina Kilbrink. Problem-based learnin g in a programming context-planning and executing a pilot survey on database access in a programming language. In George Angelos Pa- padopoulos, Wita Wojtkowski, W. Gregory Wojtkowski, Stani slaw Wrycza, and Joze Zupancic, ed- itors, Information Systems Development, Towards a Service Provis ion So...

work page 2008

[2] [5]

URLhttps://doi.org/10.48550/arXiv.2303.12961

doi: 10.48550/ARXIV .2303.12961. URLhttps://doi.org/10.48550/arXiv.2303.12961. Humza Naveed, Asad Ullah Khan, Shi Qiu, Muhammad Saqib, Saee d Anwar, Muham- mad Usman, Nick Barnes, and Ajmal Mian. A comprehensive over view of large lan- guage models. CoRR, abs/2307.06435,

arXiv doi:10.48550/arxiv.2303.12961

[3] [6]

URL https://doi.org/10.48550/arXiv.2307.06435

doi: 10.48550/ARXIV .2307.06435. URL https://doi.org/10.48550/arXiv.2307.06435. Zixuan Zhou, Xuefei Ning, Ke Hong, Tianyu Fu, Jiaming Xu, Shi yao Li, Y uming Lou, Luning Wang, Zhihang Y uan, Xiuhong Li, Shengen Y an, Guohao Dai, Xiao-Ping Zhang, Y uhan Dong, and Y u Wang. A survey on efﬁcient inference for large language models. CoRR, abs/2404.14294, 2024c. doi:

arXiv doi:10.48550/arxiv.2307.06435

[4] [9]

URL https://doi.org/10.1145/3624724

doi: 10.1145/3624724. URL https://doi.org/10.1145/3624724. Y ucheng Zhou and Guodong Long. Improving cross-modal align ment for text-guided image inpainting. In Proceedings of the 17th Conference of the European Chapter o f the Association for Computational Linguistics, pages 3445–3456, 2023a. Y ucheng Zhou and Guodong Long. Multimodal event transformer f...

doi:10.1145/3624724

[5] [10]

URLhttps://doi.org/10.48550/arXiv.2404.14294

48550/ARXIV .2404.14294. URLhttps://doi.org/10.48550/arXiv.2404.14294. Y ucheng Zhou, Zhi Rao, Jun Wan, and Jianbing Shen. Rethinkin g visual dependency in long-context reasoning for large vision-language models. arXiv preprint arXiv:2410.19732, 2024d. Yiran Zhao, Wenxuan Zhang, Guizhen Chen, Kenji Kawaguchi, a nd Lidong Bing. How do large lan- guage mode...

arXiv doi:10.48550/arxiv.2404.14294

[6] [11]

Amanda Bertsch, Maor Ivgi, Uri Alon, Jonathan Berant, Matth ew R

URL https://openreview.net/forum?id=RdJVFCHjUMI. Amanda Bertsch, Maor Ivgi, Uri Alon, Jonathan Berant, Matth ew R. Gormley, and Graham Neubig. In-context learning with long-context mode ls: An in-depth explo- ration. CoRR, abs/2405.00200,

arXiv

[7] [12]

URL https://doi.org/10.48550/arXiv.2405.00200

doi: 10.48550/ARXIV .2405.00200. URL https://doi.org/10.48550/arXiv.2405.00200. Yixing Jiang, Jeremy Irvin, Ji Hun Wang, Muhammad Ahmed Chau dhry, Jonathan H. Chen, and Andrew Y . Ng. Many-shot in-context learning in mul timodal foundation models. CoRR, abs/2405.09798,

arXiv doi:10.48550/arxiv.2405.00200

[8] [13]

URL https://doi.org/10.48550/arXiv.2405.09798

doi: 10.48550/ARXIV .2405.09798. URL https://doi.org/10.48550/arXiv.2405.09798. Yingcong Li, Muhammed Emrullah Ildiz, Dimitris S. Papailio poulos, and Samet Oymak. Transformers as algorithms: Generalization and implicit m odel selection in in-context learning. CoRR, abs/2301.07067,

arXiv doi:10.48550/arxiv.2405.09798