Zero-shot Chinese character recognition is an important topic in the signal processing field. Existing glyph-based methods achieve competitive performance by considering glyph similarity, but they incur substantial computational overhead due to the need for a visual encoder. Recognition methods based on ideographic description sequences (IDS) are more efficient, yet they do not account for inter-character similarity during representation learning, which may limit recognition performance. To alleviate this limitation, we propose a radical similarity–guided loss (RadSimLoss) for zero-shot Chinese character recognition. The RadSimLoss computes pairwise similarities between characters from their IDS, replacing glyph-based similarity, and incorporates them into the training objective. Experimental results on handwritten, scene, printed artistic, and ancient Chinese character datasets show that our proposed RadSimLoss is effective and outperforms a series of state-of-the-art approaches.

Serverless computing offers a compelling cloud model for online inference services. However, existing serverless platforms lack efficient support for GPUs, hindering their ability to deliver high-performance inference. In this article, we present Torpor , a serverless platform for GPU-efficient, low-latency inference. To enable efficient sharing of a node’s GPUs among numerous inference functions, Torpor maintains models in main memory and dynamically swaps them onto GPUs upon request arrivals (i.e., late binding with model swapping). Torpor uses various techniques, including asynchronous API redirection, GPU runtime sharing, pipelined model execution, and efficient GPU memory management, to minimize latency overhead caused by model swapping. Additionally, we design an interference-aware request scheduling algorithm that utilizes high-speed GPU interconnects to meet latency service-level objectives (SLOs) for individual inference functions. We have implemented Torpor and evaluated its performance in a production environment. Utilizing late binding and model swapping, Torpor can concurrently serve hundreds of inference functions on a worker node with 4 GPUs, while achieving latency performance comparable to native execution, where each model is cached exclusively on a GPU. Pilot deployment in a leading commercial serverless cloud shows that Torpor reduces the GPU provisioning cost by 70% and 65% for users and the platform, respectively.