Molecule graph editing has become a powerful paradigm for optimizing chemical compounds in drug discovery. Existing methods overlook the invariant structure-property relationships, and rely on variable correlations that shift across different instructions, thereby failing to generalize to out-of-distribution (O.O.D.) scenarios. To overcome the weakness of existing work, in this paper we propose to capture and utilize the invariant factors in order to achieve generalizable molecule graph editing under distribution shifts. However, this problem remains challenging, given that the invariant and variant factors are deeply entangled within the editing models. To tackle this challenge, we propose MoFE, a disentangled graph large language model for molecule graph editing that handles editing instructions under distribution shifts via disentangling invariant factors that govern editing-relevant properties. Specifically, we propose a disentangled graph projector with invariance loss that encodes molecular graphs into disentangled latent factors, with an invariance loss that ensures consistency across paraphrased prompts with the same objective. Then, we enhance the LLM with a factor-aware LoRA mixture-of-experts, where each expert is associated with a distinct latent factor. Additionally, we introduce a factor disentanglement loss weighting strategy that adaptively assigns higher weights to expert-factor pairs that perform well on relevant editing tasks. The proposed MoFE model promotes joint disentanglement between experts and latent factors, reinforcing their alignment and preventing collapse. Experiments on a representative benchmark demonstrate that MoFE is able to achieve superior O.O.D. generalization performance in molecule graph editing.

The co-design of neural network architectures, quantization precisions, and hardware accelerators offers a promising approach to achieving an optimal balance between performance and efficiency, particularly for model deployment on resource-constrained edge devices. In this work, we propose the JAQ Framework, which jointly optimizes the three critical dimensions. However, effectively automating the design process across the vast search space of those three dimensions poses significant challenges, especially when pursuing extremely low-bit quantization. Specifical, the primary challenges include: (1) Memory overhead in software-side: Low-precision quantization-aware training can lead to significant memory usage due to storing large intermediate features and latent weights for backpropagation, potentially causing memory exhaustion. (2) Search time-consuming in hardware-side: The discrete nature of hardware parameters and the complex interplay between compiler optimizations and individual operators make the accelerator search time-consuming. To address these issues, JAQ mitigates the memory overhead through a channel-wise sparse quantization (CSQ) scheme, selectively applying quantization to the most sensitive components of the model during optimization. Additionally, JAQ designs BatchTile, which employs a hardware generation network to encode all possible tiling modes, thereby speeding up the search for the optimal compiler mapping strategy. Extensive experiments demonstrate the effectiveness of JAQ, achieving approximately 7% higher Top-1 accuracy on ImageNet compared to previous methods and reducing the hardware search time per iteration to 0.15 seconds.

Drug-drug interaction (DDI) prediction is a pivotal task in biomedical research. Emerging multimodal approaches that integrate graph neural networks (GNNs) and large language models (LLMs) have gained traction, as GNNs capture molecular structures while LLMs provide a rich biomedical context. However, real-world DDI data often exhibit distribution shifts across structural and textual dimensions, stemming from variations in molecular scaffolds, drug sizes, and assay conditions. Existing methods assume an independent and identically distributed (I.I.D.) setting, failing to handle such shifts primarily due to there key limitations: (i) the entanglement of core interaction motifs with incidental structural features; (ii) inflexible message-passing GNN architectures ill-suited for diverse drug pairs; and (iii) underutilized biomedical knowledge in LLMs for capturing pairwise interaction semantics. These limitations highlight the need for a disentangled, dynamic, and pairwise-aware modeling strategy to achieve out-of-distribution generalized DDI prediction. To solve this problem, we propose DyNamic Pairwise Architecture Search for Generalizable Drug-Drug Interaction LLM (DyNAS-DDI), a novel framework that dynamically adapts network architectures for each molecular pair and integrates biomedical knowledge from LLMs to improve generalization under distribution shifts. Specifically, we propose three modules: (i) Motif-driven disentangled molecule encoding, which disentangles molecular representations into distinct motif-based features while preserving key structural signals through a self-supervised graph encoder; (ii) Attentionbased pairwise neural architecture search, where multi-head attention enriches molecular features to guide a dynamic search mechanism that adaptively optimizes message passing for diverse interaction types; and (iii) retrieval-augmented molecular instruction tuning, where external biomedical knowledge is incorporated to improve interpretability and enable reasoning for unseen drug interactions. Extensive experiments on four datasets for DDI with out-of-distribution (OOD) splits demonstrate our method's superior generalization abilities under distribution shifts. Our code can be available at https://github.com/EkkoXiao/DyNAS-DDI.

Graph neural architecture search (NAS) has achieved great success in designing architectures for graph data processing.However, distribution shifts pose great challenges for graph NAS, since the optimal searched architectures for the training graph data may fail to generalize to the unseen test graph data. The sole prior work tackles this problem by customizing architectures for each graph instance through learning graph structural information, but failed to consider data augmentation during training, which has been proven by existing works to be able to improve generalization.In this paper, we propose Data-augmented Curriculum Graph Neural Architecture Search (DCGAS), which learns an architecture customizer with good generalizability to data under distribution shifts. Specifically, we design an embedding-guided data generator, which can generate sufficient graphs for training to help the model better capture graph structural information. In addition, we design a two-factor uncertainty-based curriculum weighting strategy, which can evaluate the importance of data in enabling the model to learn key information in real-world distribution and reweight them during training. Experimental results on synthetic datasets and real datasets with distribution shifts demonstrate that our proposed method learns generalizable mappings and outperforms existing methods.

Cross-device scenarios have become increasingly common, where non-independently and identically distributed (non-IID) data is generated and stored in different devices. However, the existing cross-device NAS methods only search for a fixed architecture for different devices, neglecting that different devices have varying hardware characteristics and data distributions. In this paper, we propose a novel NAS framework that can customize the most suitable architecture for each device and its associated dataset. Specifically, we propose a decoupled data feature extractor and a device feature extractor to characterize the complex distributions of the different datasets and diverse hardware features. Then, we propose a prototype matcher to customize the operators and shape selection parameters of architectures. Experiments on ImageNet and CIFAR-10 show that our method can discover more efficient and effective architectures in cross-device scenarios than the existing approaches. To the best of our knowledge, this is the first exploration on customized cross-device NAS problem.