Shenglai Zeng, Tianqi Zheng, Chuan Tian, Dante Everaert, Yau-Shian Wang, Yupin Huang, Michael J. Morais, Rohit Patki, Jinjin Tian, Xinnan Dai, Kai Guo, Monica Xiao Cheng, Hui Liu. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Large Language Models (LLMs) enhanced with external contexts, such as through retrieval-augmented generation (RAG), often face challenges in handling imperfect evidence.They tend to over-rely on external knowledge, making them vulnerable to misleading and unhelpful contexts.To address this, we propose the concept of context-robust LLMs, which can effectively balance internal knowledge with external context, similar to human cognitive processes.Specifically, context-robust LLMs should rely on external context only when lacking internal knowledge, identify contradictions between internal and external knowledge, and disregard unhelpful contexts.To achieve this goal, we introduce Grft, a lightweight and plug-and-play gated representation fine-tuning approach.Grft consists of two key components: a gating mechanism to detect and filter problematic inputs, and low-rank representation adapters to adjust hidden representations.By training a lightweight intervention function with only 0.0004% of model size on fewer than 200 examples, Grft can effectively adapts LLMs towards context-robust behaviors.

In recent years, large language models (LLMs) have revolutionized the field of natural language processing.However, they often suffer from knowledge gaps and hallucinations.Graph retrieval-augmented generation (GraphRAG) enhances LLM reasoning by integrating structured knowledge from external graphs.However, we identify two key challenges that plague GraphRAG: (1) Retrieving noisy and irrelevant information can degrade performance and (2) Excessive reliance on external knowledge suppresses the model's intrinsic reasoning.To address these issues, we propose GraphRAG-FI (Filtering & Integration), consisting of GraphRAG-Filtering and GraphRAG-Integration. GraphRAG-Filtering employs a two-stage filtering mechanism to refine retrieved information.GraphRAG-Integration employs a logits-based selection strategy to balance external knowledge from GraphRAG with the LLM's intrinsic reasoning, reducing over-reliance on retrievals.Experiments on knowledge graph QA tasks demonstrate that GraphRAG-FI significantly improves reasoning performance across multiple backbone models, establishing a more reliable and effective GraphRAG framework.

Recent studies suggest that large language models (LLMs) possess the capability to solve graph reasoning tasks. Notably, even when graph structures are embedded within textual descriptions, LLMs can still effectively answer related questions. This raises a fundamental question: How can a decoder-only Transformer architecture understand underlying graph structures? To address this, we start with the substructure extraction task, interpreting the inner mechanisms inside the transformers and analyzing the impact of the input queries. Specifically, through both empirical results and theoretical analysis, we present Induced Substructure Filtration (ISF), a perspective that captures the substructure identification in the multi-layer transformers. We further validate the ISF process in LLMs, revealing consistent internal dynamics across layers. Building on these insights, we explore the broader capabilities of Transformers in handling diverse graph types. Specifically, we introduce the concept of thinking in substructures to efficiently extract complex composite patterns, and demonstrate that decoder-only Transformers can successfully extract substructures from attributed graphs, such as molecular graphs. Together, our findings offer a new insight on how sequence-based Transformers perform the substructure extraction task over graph data.

Graph neural networks (GNNs) have exhibited remarkable performance under the assumption that test data comes from the same distribution of training data. However, in real-world scenarios, this assumption may not always be valid. Consequently, there is a growing focus on exploring the Out-of-Distribution (OOD) problem in the context of graphs. Most existing efforts have primarily concentrated on improving graph OOD generalization from two model-agnostic perspectives: data-driven methods and strategy-based learning. However, there has been limited attention dedicated to investigating the impact of well-known GNN model architectures on graph OOD generalization, which is orthogonal to existing research. In this work, we provide the first comprehensive investigation of OOD generalization on graphs from an architecture perspective, by examining the common building blocks of modern GNNs. Through extensive experiments, we reveal that both the graph self-attention mechanism and the decoupled architecture contribute positively to graph OOD generalization. In contrast, we observe that the linear classification layer tends to compromise graph OOD generalization capability. Furthermore, we provide in-depth theoretical insights and discussions to underpin these discoveries. These insights have empowered us to develop a novel GNN backbone model, DGat, designed to harness the robust properties of both graph self-attention mechanism and the decoupled architecture. Extensive experimental results demonstrate the effectiveness of our model under graph OOD, exhibiting substantial and consistent enhancements across various training strategies. Our codes are available at https://github.com/KaiGuo20/DGAT **REMOVE 2nd URL**://github.com/KaiGuo20/DGAT.

Out-of-distribution generalization is challenging for non-participating clients of federated learning under distribution shifts. A proven strategy is to explore those invariant relationships between input and target variables, working equally well for non-participating clients. However, learning invariant relationships is often in an explicit manner from data, representation, and distribution, which violates the federated principles of privacy-preserving and limited communication. In this paper, we propose FedIIR, which implicitly learns invariant relationships from parameter for out-of-distribution generalization, adhering to the above principles. Specifically, we utilize the prediction disagreement to quantify invariant relationships and implicitly reduce it through inter-client gradient alignment. Theoretically, we demonstrate the range of non-participating clients to which FedIIR is expected to generalize and present the convergence results for FedIIR in the massively distributed with limited communication. Extensive experiments show that FedIIR significantly outperforms relevant baselines in terms of out-of-distribution generalization of federated learning.

Graph neural networks (GNNs) have received tremendous attention due to their superiority in learning node representations. These models rely on message passing and feature transformation functions to encode the structural and feature information from neighbors. However, stacking more convolutional layers significantly decreases the performance of GNNs. Most recent studies attribute this limitation to the over-smoothing issue, where node embeddings converge to indistinguishable vectors. Through a number of experimental observations, we argue that the main factor degrading the performance is the unstable forward normalization and backward gradient resulted from the improper design of the feature transformation, especially for shallow GNNs where the over-smoothing has not happened. Therefore, we propose a novel orthogonal feature transformation, named Ortho-GConv, which could generally augment the existing GNN backbones to stabilize the model training and improve the model's generalization performance. Specifically, we maintain the orthogonality of the feature transformation comprehensively from three perspectives, namely hybrid weight initialization, orthogonal transformation, and orthogonal regularization. By equipping the existing GNNs (e.g. GCN, JKNet, GCNII) with Ortho-GConv, we demonstrate the generality of the orthogonal feature transformation to enable stable training, and show its effectiveness for node and graph classification tasks.