Graph Neural Networks (GNNs) increasingly follow the ''pre-training, adaptation'' paradigm, where a GNN is pre-trained on large-scale graphs and then adapted to downstream tasks. Graph prompting adapts to the frozen encoder by modifying the input graph structure, rather than fine-tuning the model parameters. However, existing graph prompting methods often rely on probabilistic rewiring and auxiliary regularizers to control sparsity, which makes the prompting process sensitive to hyperparameters and can introduce instability in few-shot settings. To address the issue, we propose ProtoCalib, a lightweight local graph prompt for few-shot adaptation of frozen GNNs. ProtoCalib uses prototypes from the support set to score candidate edges for each anchor node. It then calibrates these scores with a per-node edge budget, which keeps the prompted graph sparse and removes the need for extra sparsity or entropy losses. We further adopt a deterministic construction of the prompted adjacency to reduce sampling noise at inference time. Extensive experiments on five graph datasets under four pre-training strategies demonstrate that our proposed ProtoCalib outshines baselines on multiple node classification datasets.

Hybrid computing-in-memory (CIM) has recently gained significant attention due to its ability to leverage the strengths of both digital CIM (DCIM) and analog CIM (ACIM). The multibit fusion (MF) scheme enhances energy efficiency by fusing low-bit results, which typically require multiple read-out cycles, into a single-cycle read out. However, the relationship between hybrid INT-CIM circuit design and network performance based on the MF scheme has not yet been systematically explored. In addition, we investigate how different MF configurations affect the performance of various neural networks. To address this gap, we first propose a less-significant group quantization (LSGQ) model, which defines and explores the design space of hybrid INT-CIM. Second, we develop a FastFuse-Search (FFS) algorithm, which optimizes configurations for different networks to strike a better balance between model accuracy and energy efficiency. Based on the experimental results, some key considerations on hybrid CIM design are derived. FFS yields a$1.72\times $energy-efficiency boost with negligible accuracy loss. Finally, we fabricate a 28-nm hybrid INT-CIM test chip, achieving 59.74 TOPS/W and 0.96 TOPS/mm2, with performance metrics of 23.21 perplexity for GPT-2, 68.69% accuracy for ResNet18, and 80.53% accuracy for ViT.