Pith. sign in

REVIEW 1 major objections 49 references

G2LoRA resolves task interference in continual learning on text-attributed graphs by projecting gradients in category-aware subspaces and modulating their magnitudes across encoders.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.3

2026-06-28 15:23 UTC pith:K7VJJ7LB

load-bearing objection G2LoRA adds category-aware projection and magnitude modulation to handle forgetting in graph-text continual learning, but the abstract leaves the actual gains hard to judge without numbers or ablations. the 1 major comments →

arxiv 2606.01873 v1 pith:K7VJJ7LB submitted 2026-06-01 cs.LG

G2LoRA: Gradient Orthogonal Low-Rank Adaptation Framework for Graph Continual Learning on Text-Attributed Graphs

Yuhan Wang , Yibo Ding , Yutong Ye , Mufan Zhao , Wenbo Zhang , Ruijie Wang , Jianxin Li This is my paper
classification cs.LG
keywords graph continual learningtext-attributed graphslow-rank adaptationcatastrophic forgettinggradient projectionmulti-modal alignmentLLM fine-tuning
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper aims to show that existing LLM-as-Aligner models suffer from catastrophic forgetting when fine-tuned sequentially on streaming tasks involving text-attributed graphs. It proposes that unifying node, link, and graph level tasks under one alignment objective allows consistent optimization. By performing category-aware gradient projection in structured subspaces, the method resolves conflicting updates and allows conditional backward transfer. Adding gradient magnitude modulation coordinates the different adaptation rates of graph and text encoders to prevent misalignment.

Core claim

G2LoRA unifies node-, link-, and graph-level tasks under a single graph-text alignment objective for consistent optimization across incremental modes. It performs category-aware gradient projection in structured subspaces to resolve conflicting updates and enable conditional backward transfer, while introducing gradient magnitude modulation to coordinate update rates between graph and text encoders.

What carries the argument

Category-aware gradient projection in structured subspaces with gradient magnitude modulation between encoders.

Load-bearing premise

That a single graph-text alignment objective produces consistent optimization across all incremental modes and that magnitude modulation alone can coordinate the differing sensitivities of the encoders.

What would settle it

An experiment where sequential fine-tuning on new tasks still causes significant degradation in alignment between graph and text embeddings or high forgetting rates on previous tasks despite using G2LoRA.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

1 major / 0 minor

Summary. The paper proposes G2LoRA, a continual learning framework for LLM-as-Aligner models on text-attributed graphs (TAGs). It unifies node-, link-, and graph-level tasks under a single graph-text alignment objective to enable consistent optimization across incremental modes. To mitigate task interference and promote positive transfer, it performs category-aware gradient projection in structured subspaces, resolving conflicting updates while enabling conditional backward transfer. Gradient magnitude modulation is introduced to coordinate update rates between graph and text encoders and prevent cross-modal drift. Experiments on benchmark datasets are claimed to show consistent outperformance over strong baselines across backbone architectures and incremental settings.

Significance. If the empirical claims hold with proper controls, the work would address a practically relevant gap in continual learning for multimodal graph models, where catastrophic forgetting and modality misalignment are common. The combination of subspace projection for interference resolution and magnitude modulation for encoder coordination offers a concrete engineering approach that could be adopted in streaming TAG applications. The unification of heterogeneous tasks under one alignment objective is a notable design choice that, if validated, simplifies the continual learning setup.

major comments (1)
  1. [Abstract] Abstract and §1: the central claims of superior continual performance, effective resolution of task interference, and balanced forward/backward transfer rest on experimental results, yet the provided text supplies no ablation studies, error bars, dataset statistics, or quantitative comparisons. This prevents evaluation of whether the proposed components deliver the claimed benefits or whether the weakest assumption (unification under a single objective plus magnitude modulation suffices without task-specific loss) holds.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the detailed review and the opportunity to clarify the presentation of our experimental validation. We address the concern point by point below.

read point-by-point responses

  1. Referee: [Abstract] Abstract and §1: the central claims of superior continual performance, effective resolution of task interference, and balanced forward/backward transfer rest on experimental results, yet the provided text supplies no ablation studies, error bars, dataset statistics, or quantitative comparisons. This prevents evaluation of whether the proposed components deliver the claimed benefits or whether the weakest assumption (unification under a single objective plus magnitude modulation suffices without task-specific loss) holds.

    Authors: We acknowledge that the abstract and §1 present high-level claims without embedding the full quantitative details. The complete manuscript contains these elements in later sections: dataset statistics and task settings appear in §3, quantitative comparisons (including forward/backward transfer metrics) are reported with error bars (std. dev. over 5 seeds) in Tables 1–4 of §4, and ablation studies isolating the category-aware projection, magnitude modulation, and single-objective unification are provided in §5 (Tables 5–7). These ablations directly compare against variants that retain task-specific losses, showing that the unified alignment objective plus our two components yields the reported gains without requiring per-task losses. If the referee’s copy omitted §4–5, we will ensure the camera-ready version cross-references key results from the abstract and §1. We can also insert a one-paragraph summary of the main quantitative findings into §1 upon revision. revision: partial

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The provided abstract and description frame G2LoRA as an engineering framework that unifies tasks under a single alignment objective, applies category-aware gradient projection in subspaces, and uses magnitude modulation for encoder coordination. No equations, fitted parameters, predictions derived from fits, or self-citations appear in the text. The central claims are presented as design choices for mitigating forgetting and interference rather than a derivation chain that reduces to its own inputs by construction. The method is self-contained against external benchmarks with no load-bearing self-referential steps.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

No free parameters, axioms, or invented entities are identifiable from the abstract alone; the work is presented as an engineering framework rather than a derivation resting on explicit postulates.

pith-pipeline@v0.9.1-grok · 5843 in / 1199 out tokens · 22911 ms · 2026-06-28T15:23:25.996965+00:00 · methodology

0 comments
read the original abstract

LLM-as-Aligner has emerged as a prevalent pre-training paradigm for Text-Attributed Graphs(TAGS), aligning graph and text modalities into a shared embedding space via CLIP-style contrastive learning. While effective on individual downstream tasks, we observe severe catastrophic forgetting when such models are sequentially fine-tuned on streaming tasks. Although parameter-efficient fine-tuning alleviates forgetting to some extent, it remains insufficient to resolve task interference and ineffective knowledge transfer. In this work, we study graph continual learning for LLM-as-Aligner models on TAGs, with the goal of mitigating interference while promoting positive transfer across tasks. This setting introduces two fundamental challenges: (1) heterogeneous downstream tasks induce shifting optimization objectives, hindering unified fine-tuning; and (2) graph and text encoders exhibit different sensitivities to adaptation, making uncoordinated updates prone to misalignment. To address these challenges, we propose G2LoRA, a continual learning framework for TAGs. G2LoRA unifies node-, link-, and graph-level tasks under a single graph--text alignment objective, and enables consistent optimization across domain/class/task incremental modes. To reduce task interference while encouraging positive transfer, G2LoRA performs category-aware gradient projection in structured subspaces, resolving conflicting updates and enabling conditional backward transfer to balance forward and backward knowledge flow. To further prevent cross-modal drift, G2LoRA introduces gradient magnitude modulation to coordinate update rates between graph and text encoders. Extensive experiments on benchmark datasets demonstrate that G2LoRA consistently outperforms strong baselines across different backbone architectures, achieving superior continual performance and transferability.

Figures

Figures reproduced from arXiv: 2606.01873 by Jianxin Li, Mufan Zhao, Ruijie Wang, Wenbo Zhang, Yibo Ding, Yuhan Wang, Yutong Ye.

Figure 1
Figure 1. Figure 1: (a) LLM-as-Aligner Performance in CL: Joint vs. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: (a) Represents the overall framework. Graph and text encoders are frozen, and only LoRA adapters are tuned. CBGP [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Left: unified downstream task. Right: unified gradi [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: (a) Performance comparison between cross-entropy [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Performance comparison of SimGCL and G2LoRA under varying K-shot configurations: (a) TIL, (b) DIL , (c) CIL. 4.8 Hyperparameter Sensitivity Analysis We analyze the sensitivity of G2LoRA to the bidirectional activa￾tion threshold 𝜖1 and the regularization coefficient 𝜆. As shown in [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Sensitivity analysis of 𝜖1 under different incremen￾tal learning settings [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

49 extracted references · 12 canonical work pages · 3 internal anchors

  1. [1]

    Hongxu Chen, Hongzhi Yin, Xiangguo Sun, Tong Chen, Bogdan Gabrys, and Katarzyna Musial. 2020. Multi-level graph convolutional networks for cross- platform anchor link prediction. InProceedings of the 26th ACM SIGKDD interna- tional conference on knowledge discovery & data mining. 1503–1511

    2020

  2. [2]

    Junru Chen, Yang Yang, Tao Yu, Yingying Fan, Xiaolong Mo, and Carl Yang

  3. [3]

    InProceedings of the 28th ACM SIGKDD conference on knowledge discovery and data mining

    Brainnet: Epileptic wave detection from seeg with hierarchical graph diffu- sion learning. InProceedings of the 28th ACM SIGKDD conference on knowledge discovery and data mining. 2741–2751

  4. [4]

    Runjin Chen, Tong Zhao, Ajay Kumar Jaiswal, Neil Shah, and Zhangyang Wang

  5. [5]

    InForty-first International Conference on Machine Learning, ICML 2024, Vienna, Austria, July 21-27, 2024

    LLaGA: Large Language and Graph Assistant. InForty-first International Conference on Machine Learning, ICML 2024, Vienna, Austria, July 21-27, 2024. OpenReview.net

    2024

  6. [6]

    Zhikai Chen, Haitao Mao, Hang Li, Wei Jin, Hongzhi Wen, Xiaochi Wei, Shuaiqiang Wang, Dawei Yin, Wenqi Fan, Hui Liu, et al . 2024. Exploring the potential of large language models (llms) in learning on graphs.ACM SIGKDD Explorations Newsletter25, 2 (2024), 42–61

    2024

  7. [7]

    Ziyang Cheng, Zhixun Li, Yuhan Li, Yixin Song, Kangyi Zhao, Dawei Cheng, Jia Li, Hong Cheng, and Jeffrey Xu Yu. 2025. Can LLMs Alleviate Catastrophic Forgetting in Graph Continual Learning? A Systematic Study.arXiv preprint arXiv:2505.18697(2025)

    work page arXiv 2025

  8. [8]

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. Bert: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 conference of the North American chapter of the association for computational linguistics: human language technologies, volume 1 (long and short papers). 4171–4186

    2019

  9. [9]

    Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, et al. 2024. The llama 3 herd of models.arXiv preprint arXiv:2407.21783 (2024)

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  10. [10]

    Will Hamilton, Zhitao Ying, and Jure Leskovec. 2017. Inductive representation learning on large graphs.Advances in neural information processing systems30 (2017)

    2017

  11. [12]

    Xiaoxin He, Xavier Bresson, Thomas Laurent, Adam Perold, Yann LeCun, and Bryan Hooi. 2023. Harnessing explanations: Llm-to-lm interpreter for enhanced text-attributed graph representation learning.arXiv preprint arXiv:2305.19523 (2023)

    work page arXiv 2023

  12. [13]

    Thanh Duc Hoang, Do Viet Tung, Duy-Hung Nguyen, Bao-Sinh Nguyen, Huy Hoang Nguyen, and Hung Le. 2023. Universal graph continual learning. arXiv preprint arXiv:2308.13982(2023)

    work page arXiv 2023

  13. [14]

    Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, Weizhu Chen, et al. 2022. Lora: Low-rank adaptation of large language models.ICLR1, 2 (2022), 3

    2022

  14. [15]

    Weihua Hu, Matthias Fey, Marinka Zitnik, Yuxiao Dong, Hongyu Ren, Bowen Liu, Michele Catasta, and Jure Leskovec. 2020. Open graph benchmark: Datasets for machine learning on graphs.Advances in neural information processing systems 33 (2020), 22118–22133

    2020

  15. [16]

    TN Kipf. 2016. Semi-supervised classification with graph convolutional networks. arXiv preprint arXiv:1609.02907(2016)

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  16. [17]

    James Kirkpatrick, Razvan Pascanu, Neil Rabinowitz, Joel Veness, Guillaume Desjardins, Andrei A Rusu, Kieran Milan, John Quan, Tiago Ramalho, Agnieszka Grabska-Barwinska, et al. 2017. Overcoming catastrophic forgetting in neural networks.Proceedings of the national academy of sciences114, 13 (2017), 3521– 3526

    2017

  17. [18]

    Zhizhong Li and Derek Hoiem. 2017. Learning without forgetting.IEEE transac- tions on pattern analysis and machine intelligence40, 12 (2017), 2935–2947

    2017

  18. [19]

    Yan-Shuo Liang and Wu-Jun Li. 2024. InfLoRA: Interference-Free Low-Rank Adaptation for Continual Learning. InIEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2024, Seattle, W A, USA, June 16-22, 2024. IEEE, 23638–23647

    2024

  19. [20]

    Hao Liu, Jiarui Feng, Lecheng Kong, Ningyue Liang, Dacheng Tao, Yixin Chen, and Muhan Zhang. 2023. One for all: Towards training one graph model for all classification tasks.arXiv preprint arXiv:2310.00149(2023)

    work page arXiv 2023

  20. [21]

    Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. 2019. Roberta: A robustly optimized bert pretraining approach.arXiv preprint arXiv:1907.11692 (2019)

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  21. [22]

    Yuhang Liu, Minglai Shao, Zengyi Wo, Yunlong Chu, Bing Hao, Shengzhong Liu, Ruijie Wang, and Jianxin Li. 2025. Learning Noise-Resilient and Transferable Graph-Text Alignment via Dynamic Quality Assessment. arXiv:2510.19384

    work page arXiv 2025

  22. [23]

    Zheyuan Liu, Xiaoxin He, Yijun Tian, and Nitesh V. Chawla. 2024. Can we Soft Prompt LLMs for Graph Learning Tasks?. InCompanion Proceedings of the ACM on Web Conference 2024, WWW 2024, Singapore, Singapore, May 13-17, 2024, Tat- Seng Chua, Chong-Wah Ngo, Roy Ka-Wei Lee, Ravi Kumar, and Hady W. Lauw (Eds.). ACM, 481–484

    2024

  23. [24]

    Zemin Liu, Xingtong Yu, Yuan Fang, and Xinming Zhang. 2023. Graphprompt: Unifying pre-training and downstream tasks for graph neural networks. InPro- ceedings of the ACM web conference 2023. 417–428

    2023

  24. [25]

    Chaoxi Niu, Guansong Pang, and Ling Chen. 2024. Graph continual learning with debiased lossless memory replay.arXiv preprint arXiv:2404.10984(2024)

    work page arXiv 2024

  25. [26]

    Chaoxi Niu, Guansong Pang, Ling Chen, and Bing Liu. 2024. Replay-and-forget- free graph class-incremental learning: A task profiling and prompting approach. Advances in Neural Information Processing Systems37 (2024), 87978–88002

    2024

  26. [27]

    Xiaokang Peng, Yake Wei, Andong Deng, Dong Wang, and Di Hu. 2022. Balanced multimodal learning via on-the-fly gradient modulation. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 8238–8247

    2022

  27. [28]

    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. InInternational conference on machine learning. 8748–8763

    2021

  28. [29]

    Gobinda Saha, Isha Garg, and Kaushik Roy. 2021. Gradient Projection Memory for Continual Learning. In9th International Conference on Learning Representations, ICLR 2021, Virtual Event, Austria, May 3-7, 2021. OpenReview.net

    2021

  29. [30]

    Prithviraj Sen, Galileo Namata, Mustafa Bilgic, Lise Getoor, Brian Galligher, and Tina Eliassi-Rad. 2008. Collective classification in network data.AI magazine29, 3 (2008), 93–93

    2008

  30. [31]

    Xiangguo Sun, Hong Cheng, Jia Li, Bo Liu, and Jihong Guan. 2023. All in one: Multi-task prompting for graph neural networks. InProceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 2120–2131

    2023

  31. [32]

    Jiabin Tang, Yuhao Yang, Wei Wei, Lei Shi, Lixin Su, Suqi Cheng, Dawei Yin, and Chao Huang. 2024. Graphgpt: Graph instruction tuning for large language models. InProceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval. 491–500

    2024

  32. [33]

    Qi-Wei Wang, Da-Wei Zhou, Yi-Kai Zhang, De-Chuan Zhan, and Han-Jia Ye

  33. [34]

    Advances in Neural Information Processing Systems36 (2023), 15060–15076

    Few-shot class-incremental learning via training-free prototype calibration. Advances in Neural Information Processing Systems36 (2023), 15060–15076

    2023

  34. [35]

    Ruijie Wang, Zheng Li, Dachun Sun, Shengzhong Liu, Jinning Li, Bing Yin, and Tarek Abdelzaher. 2022. Learning to sample and aggregate: few-shot reason- ing over temporal knowledge graphs. InProceedings of the 36th International Conference on Neural Information Processing Systems (NeurIPS ’22)

    2022

  35. [36]

    Ruijie Wang, Zheng Li, Jingfeng Yang, Tianyu Cao, Chao Zhang, Bing Yin, and Tarek Abdelzaher. 2023. Mutually-paced Knowledge Distillation for Cross-lingual Temporal Knowledge Graph Reasoning. InProceedings of the ACM Web Conference 2023 (WWW ’23)

    2023

  36. [37]

    Ruijie Wang, Yutong Zhang, Jinyang Li, Shengzhong Liu, Dachun Sun, Tianchen Wang, Tianshi Wang, Yizhuo Chen, Denizhan Kara, and Tarek Abdelzaher. 2024. MetaHKG: Meta Hyperbolic Learning for Few-shot Temporal Reasoning. In Proceedings of the 47th International Conference on Research and Development in Information Retrieval (SIGIR ’24)

    2024

  37. [38]

    Zhihao Wen and Yuan Fang. 2023. Augmenting Low-Resource Text Classifica- tion with Graph-Grounded Pre-training and Prompting. InProceedings of the 46th International ACM SIGIR Conference on Research and Development in Infor- mation Retrieval, SIGIR 2023, Taipei, Taiwan, July 23-27, 2023, Hsin-Hsi Chen, Wei-Jou (Edward) Duh, Hen-Hsen Huang, Makoto P. Kato,...

    2023

  38. [39]

    Nan Wu, Stanislaw Jastrzebski, Kyunghyun Cho, and Krzysztof J Geras. 2022. Characterizing and overcoming the greedy nature of learning in multi-modal deep neural networks. InInternational Conference on Machine Learning. 24043–24055

    2022

  39. [40]

    Xixi Wu, Yifei Shen, Fangzhou Ge, Caihua Shan, Yizhu Jiao, Xiangguo Sun, and Hong Cheng. 2025. A Comprehensive Analysis on LLM-based Node Classification Algorithms.CoRRabs/2502.00829 (2025)

    work page arXiv 2025

  40. [41]

    Yichen Wu, Hongming Piao, Long-Kai Huang, Renzhen Wang, Wanhua Li, Hanspeter Pfister, Deyu Meng, Kede Ma, and Ying Wei. 2025. Sd-lora: Scal- able decoupled low-rank adaptation for class incremental learning.arXiv preprint arXiv:2501.13198(2025)

    work page arXiv 2025

  41. [42]

    Hao Yan, Chaozhuo Li, Ruosong Long, Chao Yan, Jianan Zhao, Wenwen Zhuang, Jun Yin, Peiyan Zhang, Weihao Han, Hao Sun, et al. 2023. A comprehensive study on text-attributed graphs: Benchmarking and rethinking.Advances in Neural Information Processing Systems36 (2023), 17238–17264

    2023

  42. [43]

    Guanxiong Zeng, Yang Chen, Bo Cui, and Shan Yu. 2019. Continual learning of context-dependent processing in neural networks.Nature Machine Intelligence1, 8 (2019), 364–372

    2019

  43. [44]

    Peiyan Zhang, Yuchen Yan, Chaozhuo Li, Senzhang Wang, Xing Xie, Guojie Song, and Sunghun Kim. 2023. Continual learning on dynamic graphs via parameter isolation. InProceedings of the 46th international ACM SIGIR conference on research and development in information retrieval. 601–611

    2023

  44. [45]

    Xikun Zhang, Dongjin Song, and Dacheng Tao. 2022. Hierarchical prototype networks for continual graph representation learning.IEEE Transactions on Pattern Analysis and Machine Intelligence45, 4 (2022), 4622–4636

    2022

  45. [46]

    Xikun Zhang, Dongjin Song, and Dacheng Tao. 2023. Ricci curvature-based graph sparsification for continual graph representation learning.IEEE Transactions on Neural Networks and Learning Systems(2023). Gradient Orthogonal Low-Rank Adaptation Framework for Graph Continual Learning KDD ’26, August 09–13, 2026, Jeju Island, Republic of Korea

    2023

  46. [47]

    Xikun Zhang, Dongjin Song, and Dacheng Tao. 2024. Continual learning on graphs: Challenges, solutions, and opportunities.arXiv preprint arXiv:2402.11565 (2024)

    work page arXiv 2024

  47. [48]

    Da-Wei Zhou, Zi-Wen Cai, Han-Jia Ye, De-Chuan Zhan, and Ziwei Liu. 2025. Revisiting class-incremental learning with pre-trained models: Generalizability and adaptivity are all you need.International Journal of Computer Vision133, 3 (2025), 1012–1032

    2025

  48. [49]

    Yun Zhu, Haizhou Shi, Xiaotang Wang, Yongchao Liu, Yaoke Wang, Boci Peng, Chuntao Hong, and Siliang Tang. 2025. Graphclip: Enhancing transferability in graph foundation models for text-attributed graphs. InProceedings of the ACM on Web Conference 2025. 2183–2197

    2025

  49. [50]

    Yun Zhu, Yaoke Wang, Haizhou Shi, and Siliang Tang. 2024. Efficient Tuning and Inference for Large Language Models on Textual Graphs. InProceedings of the Thirty-Third International Joint Conference on Artificial Intelligence, IJCAI 2024, Jeju, South Korea, August 3-9, 2024. ijcai.org, 5734–5742. A Appendix A.1 Proof of Theorem 1 (1) For Rule 1, we have 𝑊...

    2024